Infosys | Microsoft

     We are witnessing information explosion over the internet, tons of information is getting accumulated in. However, we still struggle to get the “Accurate and Authentic Data”. Have you ever needed the zip code of a city, route to reach a place, dining menu of restaurants, weather forecast and history, crime rates in a specific area of city? This list just goes on. How do we get this data? Search this information on our favorite search engines and rest in peace when we find it!! But, do we really know whether the data which we got is really accurate?! It could be stale, misleading or just plain inaccurate!! Why can’t you get information as easily as you can get a size 40 Creamy white Louis Philippe shirt or maybe a striking green 8 GB IPOD shuffle; because INFORMATION is not commodity yet!! 

     In contrast you may come across a lot of mash up applications which consume data from genuine sources and present it to the end user in his context. This data is not only reliable but also closely up to date in terms of its accuracy.  The specialized organizations or vendors, who are responsible to publish data in some predefined format so that it can be consumed by a variety of applications, are called as Content Providers. Content providers expose their data in the form of feeds or messages abiding by some standards so that the consumer can easily make use of the data.
     So now we know who can give us accurate and authentic data. Next, we need to find content providers who are specialized in providing the contents to be consumed in our applications. This data is so important for applications or businesses that, we are ready to pay for the usage of the data but the challenge is to filter through numerous content provider sites again via the search engine to look for the appropriate ones. Why can’t we have something or someplace which is like a one stop shop for all popular Content Providers?
Well our prayers have been answered!! Ever since Microsoft has announced Azure Services, they are updating and trying hard to make their offerings richer. In PDC 2009, Microsoft has announced a project code named “Dallas” which addresses all the above pain points for businesses. In the world of X as a Service, you can call out “Project Dallas” with “Information as a Service” but in the true sense it is much more than that. Microsoft made the announcement about Project Dallas in PDC 2009 and further details can be found here introducing Dallas.
     Project Dallas is an information market place; yes it is a market place. Dallas brings information together into a single location. Be this information in any form like images, documents, real time data etc. Unified provisioning and billing framework are very important characteristics of this information market place. Dallas provides required setup for a market place and brings together the content providers and the consumers. As most of us have desired for to trade information like a commodity, it seems to be here. Discovery is another key distinctive of any market place.  Dallas provides the discovery services to be able to find content providers for a specific business domain. Contents can be viewed on the portal itself to get a quick snapshot of the information.
At this point let me point out another issue in our search for information. Once we finalize on the Content Provider we need to evaluate the interoperability efforts and the plumbing to be done in our application to consume the data. And what if we want to change or add a new content provider? Do we have to go back to do the plumbing again?  Sounds very cumbersome, right? Looks like Microsoft has been thoughtful enough to sort this aspect as well. Dallas has very simple provisioning process where a consumer needs to discover and select the provider, take a unified view of the data on Dallas itself and start using and benefitting from it. Dallas supports an inbuilt billing framework to provide consumption details to consumer as well as revenue details to the content providers.  So bye bye cumbersome and proprietary plumbing!!
     From the business user’s point of view, the dataset from the content providers are rendered within the Dallas portal along with a very powerful analytics tool “Power Pivot”.  As an independent analyst, you buy a subscription, use power pivot and take home what you need. It’s that simple. Dallas provides capabilities to bring disparate data at one place to slice and dice the data, analyze it eventually empowering you to make decisions. Analytics capabilities are extended to consume these contents from within Excel using Power Pivot, Microsoft Office and SQL Server for rich reporting and analytics. Your transactional data combined with Reference data brought from Dallas gives broader data points for analysis.

Dallas - High Level Architecture

*Source: Microsoft PDC 2009 Presentation on Project Dallas

     The most important part of “Project Dallas” is its integration capability in application development. Once you select the content provider, Dallas would provide you a proxy class to consume the contents. Download this proxy class refer it in your code and in three lines of code in C#; you are ready to use the information from the content providers. That’s it!! With the same ease you can shift to any other content provider available at “Dallas” market place, so no worries of locking and reinvesting efforts to integrate with other content providers. Doesn’t it sound like getting the information as a commodity and use it at our ease. Technically- since the dataset format received through proxy class from Dallas content partner is similar, only proxy class should change and further implementation should not be affected much. Till now the proxy class is available only in C# and it shouldn’t be a problem to provide it for popular programming languages and frameworks. Dallas is a cross platform server hence the contents from Dallas can bring the disparate data together to widen the analysis measures on varied platforms like Windows, Apple iPhones etc.
     Microsoft hasn’t forgotten the Content Providers. If you as a Content Provider have premium content, Dallas is the place to go and publish your contents. Dallas provides content partnering business opportunities to businesses to publish their contents, make it available to consumers and offload the licensing and billing management to Dallas. Content providers have to register themselves as partners, publish their data as web service and they are ready with their business.  It appears to be a quick setup business model for content providers. 
    Come Dallas and we can very easily develop creative mash up applications. Imagine applications mashing contents like maps, weather forecast and history, crime rates, new business, news feeds, reviews, real estate contents, Business relevant feeds from research firms ….. And lot more.  On the other hand you can use these contents to form a new and enriched analytical measure on your business data. Dallas is going to ”COMMODITIZE INFORMATION”, which can be consumed independently for better impact. Combined with your own transactional data to this is definitely going to raise analysis to a new level so that hitherto tricky questions like-  How are transactions impacted because of crime rate in a neighborhood or What are the reviews of your store with respect to service under a specific manager’s shift will be answered with great ease!!

Posted by Sachin Prakash Sancheti at 10:06 AM | Permalink | Comments (1) | TrackBacks (0)

The Branch site/PoP site administration always had its own contentious points on its RoI and efficiency. So far there was no inherent support from MS on this, except GPO based throttling and BITS. With Win7 and server 2008, MS proves its focus on helping enterprises with cost-effective and efficient solutions for branch administration to increase the user productivity.

Branch caching is a new feature available in Windows 7 ideal for a typical branch office. It is integrated with BITS and is capable of caching http, https and SMB. From a security standpoint, it works seamlessly with SSL, IPSec and SMB signing. Also the solution is flexible for branch offices with/without local servers. Branch Caching should be looked upon as a feature which can complement the existing infrastructure. It brings in more value when used in conjunction with technologies like DFS, SCCM distribution, web servers etc. It is no surprise that many of the early adopters for Win 7 are looking at branch caching as a prospective solution for their branch user computing.

Together with windows server 2008, enterprises can provide more efficient solutions to their branch office users. The new active domain service in server 2008 supports read only domain controllers (RoDC) which can be of great use in offices with limited physical security. The addition of RoDC in a branch environment can substantially improve the user experience through faster authentication and authorization. Windows Server 2008 R2 comes with significant improvements in TCP/IP stack along with SMB 2.0 and other enhancements in file systems. These changes are focused on improving the efficiency and reducing the latencies, delays and utilization on WAN links. ------ Yogesh K G - Windows 7 / Windows server 2008 Architect

Posted by Saju Sankarankutty at 7:22 AM | Permalink | Comments (0) | TrackBacks (0)

For all these years if a .Net architect has to model the software system. He or she has to rely on modeling tools like Rational XDE, or Visio Enterprise Architect. Personally I was never impressed with Visio Enterprise architect’s modeling support, and code generation it has to offer. Though Visio has several stencils, templates, symbols available; UML modeling and associated code generation was always bit stiff. Integration with Visual studio to synch up models with code and vice a versa was another challenge.  3^rd party tools like Rational XDE has good support for .NET but one has to pay hefty license fees to use such tools.

Result of this, system modeling used to get constrained into Microsoft Word, Power Points, Visio’s. Keeping Word/Visio based models up to date w.r.to the architecture, design, code changes was always a catching game leading the code, designs and overall system documentation out of synch impacting traceability between these artifacts.

Increasingly this has caused the disharmony between architecture modeling and development teams.

After all these years, finally, Microsoft Visual Studio 2010(VS2010) seem to have helped overcome this obstacle by embracing Unified Modeling Language (UML) and making architecture, design, development, testing seamlessly possible through its Integrated development environment. If not all, VS2010 has support for most of the UML diagramming types like

1.       Use case diagram

2.       Activity diagram

3.       Sequence diagram

4.       Class diagram

5.       Component diagram

With VS2010 beta 2, while there are a few UML diagram types i found missing at this point of time like the ones of State life cycle, object, deployment etc. I was impressed with the inclusion of diagram type called as “Layered diagram”. Through this modeling diagram, one can not only depict various logical layers like Presentation (UI), Business logic, Data Access, Utility and Database but also define the dependencies/coupling between them in terms which layer can call which one. This helps in establishing and adhering the architecture rules set by an architect while architecting the system like Presentation layer components should not directly call the data access layer component and vice a versa.

To validate the architecture in code, one can map various project (UI, Service, Data access) files by dragging and dropping into this layer diagram in respective layers and then validate it by using VS2010 “Validate Architecture” option which tracks down any violations in the code.

So far while most of the architects used to write similar rules as part of MS word architecture document but it was left with the wisdom of developer to adhere or violate it, catching such architecture principle violations were difficult in the million Lines of code developed.

Through Layer diagramming architecture validation support, VS2010 will help in keeping the design and code in synch with the architecture and design principles established during the early stages of life cycle. However this is not all with VS2010 architecture modeling support, I have just scratched the surface of what VS2010 has on offer and intend to cover other aspects in coming blogs.

Posted by Sudhanshu Hate at 11:42 AM | Permalink | Comments (0) | TrackBacks (0)

Few weeks back we got access to HP's Touch Smart laptop with Win 7 and that was our first real exposure to working with Touch. Touch isn't something of a science fiction anymore and most new mobile devices support touch capabilities, though not necessarily multi-touch. Microsoft Surface  was a very interesting innovation in this space, but was limited to some extent on its usage due to factors like restricted availability of surface table, its cost, and it being horizontal and hence not suitable for many business applications.

Windows 7 now natively supports multi-touch and if you have the right hardware like HP touch smart laptop, you can start to program some real cool applications. When I first read about WPF 4.0 will support touch at framework level, I wasn't sure what it meant and hence I decided to spend some time understanding this a bit more. Here I am not going deep into any of these aspects, but just want to highlight a few important aspects of touch and how to program for it.

The first question really is what is multi-touch and how is this different from say the stylus based Tablet PCs. Simplistically put, multi-touch literally means ability to handle multiple touch points at the same time. So if you can put more than 1 finger on the device, it is able to recognize all of them and provide these via relevant APIs for programmatic manipulation. To add to this will be ability to recognize gestures like pan, zoom, rotate etc and again providing suitable API backing for applications to program against them.

The next question that needs some digging into is, do I really need to handle any specific messages/API? When I run existing applications on Win 7 touch device, I am able to get basic manipulation working without having done anything. The point to understand here is that the underlying drivers and OS do provide for touch specific information, but since many applications (or rather almost all applications as of today) aren't touch aware, they still need to be responsive in some sense. Hence the touch messages are eventually promoted as regular mouse messages. It is due to this behavior that most applications will continue to work without any specific handling of touch. However the response may not be very good and at times seem jerky. Also you may end up with unwanted behavior like in one of our apps, we had a touch and move behavior, but while moving, the finger would go over another control and it would react even though it wasn't supposed to. The mouse event promotion probably triggered the other control to also react.

To address this an important aspect is ability the suspend mouse event promotion. The high level behavior hence will be to capture touch down event, suspend mouse event promotion, continue to handle touch related events to provide for pan, zoom, flip, rotate etc and finally handle touch up to revert to normal behavior. In our application we did this and we had much better touch response from the application (apart from the hardware issues where the touch calibration will frequently go out of synch).

Applications like IE 8 and Paint are already touch aware and if using them on Win 7 with touch hardware, you can play around with them. IE will allow zoom in-out of content via touch and paint will allow you to draw with multiple fingers. To know more about touch support, check these links

1. Windows Touch
2. Touch support in Silverlight
3. WM_TOUCH
4. WM_GESTURE

Finally, one last aspect is do I need to do something different when designing for touch aware applications. One answer to this is already discussed in previous question in terms of WM_TOUCH message handling. Some of the other aspects are to create larger controls so that touching is easier and the application can better respond to it. Also note certain behaviors will also have to change like you can no longer program for mouse hover and tooltip kind of user experience as these won't work well with touch interface. When using stylus, it does offer proximity sensor and when it is close to the screen, it can behave like mouse hover, but not with regular touch. You can check out more design guidance on touch applications here. Another document of interest will be How to Design and Test Multitouch Hardware Solutions for Windows 7.

The various links that I have provided above should help you get started on touch journey. There is also an interesting session from PDC 2009 on Windows Touch Deep Dive. However note that this talks at low level details and raw windows API and programming and isn't really talking about managed programming.

Hope you will find this interesting and would like to get started with touch interface is not already doing that. I would like to hear from you in terms of would you consider touch in your applications and what kind of applications would they be?

Posted by Atul Gupta at 8:09 AM | Permalink | Comments (3) | TrackBacks (0)

Yes, this title is influenced by death by chocolate, where in you get an overdose of chocolate. At this time I feel exactly the same for Silverlight (SL). With just over 2 years since the first version made its mark felt, Silverlight has come a long long way. Ironically, as part of TechDays event hosted specifically at our campus, we talked a lot about SL3 and right then, across the ocean, at PDC 2009, Microsoft just unveiled SL4 beta bits.

When we started looking at SL 1.0 back in late 2007, it had limited feature set with XAML support, and most work had to be done in java script. It looked more of media (video) playback at that time. MS called it their RIA platform, but it didn't offer much at that time. With SL2 at PDC 2008 and SL3 just earlier this year in July 2009, a host of features have found their way in the platform. Along with multitude of controls, to .net language support, to IIS Smooth streaming, to perspective 3D, to out of browser experience, SL is a technology you just cannot ignore.

It was very interesting to note that since keeping the size of the plugin small was critical part of SL implementation, MS actually ended up reducing the size of the SL 3 plugin over its predecessor, SL2. I haven't yet checked on the size for SL4 and given that it is still in beta it also won't make sense to do it right now. Anyway, so MS announced the public beta of SL 4 at PDC 2009 and the new developer tool set is available here. Interestingly, SL4 beta will work only with VS 2010 Beta 2 for now.

You can check this paper to get overview on what are the new features in SL4 and as well this session by Karen Corby at PDC 09. Here i will also give a quick snapshot of features I found interesting. These aren't necessarily on the only new features and may also not fall high in your own priority listing.

Few SL4 Features

1. Print support is finally here. This is one requirement that we were asked a lot about
2. Media support now also includes ability to capture devices like webcams and stream video as recorded by it
3. More support for writing business apps with things like clip board access, rich text box support, bi-directional support to handle languages like Arabic, NGEN for platform assemblies to provide faster load times
4. Styles can now be implicitly applied to all same type controls, just like WPF and you don't need to always use a Key
5. Drag and drop from desktop onto SL application is supported
6. Right click menu can now be customized

As I said, there are many more features to keep a watch on, but these are just high level ones that caught my attention. There are many more videos worth watching from PDC, in case you didn't get to attend the event in person.

Finally, When will RTM for SL4 be available? In his keynote address on Day 2 of PDC 2009, Scott Guthrie had said that final release will ship in the first half of next year. My personal take on this will be along with VS 2010 in March 2010 at MIX 2010 [Disclaimer: This purely a personal opinion and not an official statement].

Posted by Atul Gupta at 10:12 AM | Permalink | Comments (2) | TrackBacks (0)

Recently during one of the client discussions, I was asked to summarize 5 key points in 1 minute on why the client should entrust technology refresh program to us. Suddenly caught up in this situation, this is what I described:

1. The proposition of Infosys Migration Framework to migrate technology platforms with minimum business interruptions & downtime coupled with savings of $90- $160 per PC per annum (based on the study and assessment from Microsoft) is very compelling.

2. Besides being a Platinum Sponsor for the recently concluded Microsoft New Efficiency Launch (Windows 7, Windows 2008 R2 and Exchange 2010); Infosys has been part of Windows 7 Technology adoption program (TAP) in the capacity of the systems integrator and by extensive product testing has developed strong expertise on the product to help organizations to exploit the full potential of Windows 7 in enterprise wide rollout.

3. Infosys as an early adopter of Windows 7, has done pilot on 200 machines and plan to move 70% of our machines to Windows 7.

4. We are currently helping some of the large enterprises across geographies and industries for Windows 7 Pilot / POC, readiness assessments and actual migrations.

5. With a breadth of solution offerings, accelerators, strategies, best practices from current client engagements and past experience, Infosys offers a worry free migration to Windows 7 with expected reduction of migration cost upto 35%.

When I think of this again, obviously there are many aspects which differntiates us and help our clients in their transformation journey... watch this space for more on those...

Posted by Rajesh Gupta at 8:07 PM | Permalink | Comments (0) | TrackBacks (0)

I knew since this was printer related it had to be somewhere in the "Devices and Printers" dialog but opening it up, I could not see this option. I tried directly searching from the start menu but no luck. Right click on a printer only allowed to set it as default printer.

However while trying that I could suddenly see the option "Manage default printers" on the blue bar in the "Devices and Printer" dialog. That is when I realized that this option is visible only when you select a printer in the list of printers already available. See the picture below that shows this menu option.

Not seeing the option when I don't have any printer installed makes sense, but having to select one in the list before I could set the default printers as per locations is a bit surprising. Later I found more details around this option here.

Posted by Atul Gupta at 6:15 AM | Permalink | Comments (0) | TrackBacks (0)

In this blog we’ll see how e-mail notifications can be sent to the participants to cast their vote for a poll using Microsoft live.

1. To send e-mails to the participants with the poll details, MYOC sends it through live server. It requires a valid live id to send e-mails. Below are the configuration settings used -

2. Create the message to be sent, get the e-mail IDs of the participants and write the code below to send e-mails from a specified address.

In this way e-mail notifications can be sent to participants.

Posted by Divya Sharma at 1:12 PM | Permalink | Comments (0) | TrackBacks (0)

In this blog we’ll see in details how sending a tweet from a particular twitter account programmatically works on Azure.

1. Take twitter ID and password as input from user, encrypt it and generate twitter authorization key from it. This key will be used to authorize a twitter user when calling the service to send a tweet.
string authorizationKey = Convert.ToBase64String(System.Text.Encoding.ASCII.GetBytes(
                              twitterUserId + ":" +
                              twitterPwd));

2. Create the status message to be sent on the twitter account
string messageToSend = "Hi";

// the message has to be enclosed in between opening <status> and closing </status> tags
// if the message length is greater than 140, the message will be trimmed to 140 characters
// (including white spaces)
string status = "<status>" + messageToSend + "</status>";

3. Use the below mentioned twitter service URI for request
// this is the URI of twitter service to update status of a twitter account
Uri serviceUri = new Uri(@"http://twitter.com/statuses/update.xml");

4. Set the Expect100Continue property as false before sending the request to twitter service. Read the post to know why is it required.
System.Net.ServicePointManager.Expect100Continue = false;

5. Now create the request and add parameters to the request. The request body is encoded using UTF 8 encoding. There are saveral other methods to encode which provide encryption too. Please check this link for more details.
WebHeaderCollection header = new WebHeaderCollection();

// add twitter authorization key to request header
header.Add("Authorization", "Basic " + twitterAuthorizationKey);

// create request and add header
HttpWebRequest request = WebRequest.Create(serviceUri) as HttpWebRequest;
request.Headers = header;

// encode the request body
byte[] requestBodyBytes = Encoding.UTF8.GetBytes(status);

//add request parameters
request.ContentLength = requestBodyBytes.Length;
request.ContentType = "application/xml";
request.Method = "POST";

//create request stream to send to the twitter service
Stream requestStream = request.GetRequestStream();
try
{
      requestStream.Write(requestBodyBytes, 0, requestBodyBytes.Length);
}
finally
{
      requestStream.Close();
}
     
6. Post the request to the twitter service URI and get the response
using (HttpWebResponse response = request.GetResponse() as HttpWebResponse)
{
      Stream stream = response.GetResponseStream();
using (StreamReader reader = new StreamReader(stream))
      {
             string contents = reader.ReadToEnd();
             stream.Close();
      }
}

Following the above approach, one can send a tweet on his/her twitter account, which his/her followers can see in their updates.

Posted by Divya Sharma at 1:03 PM | Permalink | Comments (0) | TrackBacks (0)

MYOC (Make Your Opinion Count) – an online poll application hosted on Microsoft Azure, uses Twilio to make it easier for people to participate in the online polls. Twilio is telephony in the cloud which exposes RESTful APIs to build scalable voice applications. It supports both inbound and outbound telephony calls. Pricing is developer friendly with pay-as-you-go model.

MYOC uses Twilio in two ways –
1. Poll creator can place a call  for participant to caste his/her vote
2. A participant can dial-in for a particular poll to caste his/her vote

Let’s see what all it takes to use twilio in MYOC to call a participant and accept his/her vote or to handle an incoming call to cast a vote.

Twilio Account Setup

1. The first and foremost thing is to have a twilio account. To play around, you can register for a free account which gives $30 credit. Open the link https://www.twilio.com/try-twilio, provide your details and click on ‘TRY TWILIO’ button.
 

2. This will create a user account on twilio and show account dashboard. It will display following things –
a. API Credentials – which has account Sid and auth token to be used while calling twilio REST APIs from application
 

b. Trial Sandbox Details
 

c. Account Balance – it will show current available account balance

Getting Ready
1. Download twilio documentation from http://www.twilio.com/resources/tarball/twilio-documentation.zip to know how to use twilio in an application.

2. Download C# helper libraries for twilio from http://www.twilio.com/docs/libraries and add twiliorest.cs in the webrole project.

3. Add the parameters in csdef file as below -

4. Add the values of above defined settings in cscfg file as below -

Placing an outbound call on a given number

1. Read values form configuration file

2. Initialize call process by creating twilio account instance with account SID & auth token and creating the service URL to be called once participant receives the call.

3. Place the call on specified contact number from caller ID.

4. The service invoked with the pollURL will handle the user response over call. User can provide the dialogues to use in the call and what action to take according to user input using Twilio Markup Language(TwiML). TwiML contains tags to define various activities like say a text, pause for defined time, get user input, play a recorded file, record the voice, call another service etc. Please refer to the TwiML documentation available at http://www.twilio.com/docs/api/2008-08-01/twiml/ to get more details. The service to handle the call will be as shown below -

Handling an Inbound Call

1. Map the ‘Uses URL’ of the twilio account with the application service URL and save. When user dials in the twilio number, this service will be invoked.

2. When participant logs in MYOC to vote for a poll, he/she will be given the dial-in number with PIN number and the poll ID. Participant will call the given number and enter the PIN, then he/she will be redirected to the service specified in Uses URL of twilio account.

3. The service will be similar to the service used in placing an outbound call using twilio. The only difference will be that participant will be prompted for entering the poll ID for which he/she wishes to vote.

4. The service to handle the call will be as shown below –

5. If the participant has entered a valid poll ID then present the poll choices to the user and get his/her response and submit his vote as explained in the step 4 of above section ‘Placing an outbound call on a given number’.

Posted by Divya Sharma at 12:16 PM | Permalink | Comments (0) | TrackBacks (0)

Generally when it comes to migrating heavy data from one SQL server to another, among others following approaches are widely used:

1.       Using SSIS package

2.       And using BCP.

One needs to keep in mind a few points for the fruitful migration without any anomaly.

For one of our project we used SSIS package for migrating data but hit upon a few glitches in the course of our task. One needs to keep in mind the following points:

1. Since we will be running the SSIS generally on-premise to migrate data located on-premise to cloud we need to make sure the port number 1433 is opened for ctp.database.windows.net (or please check for the recent dns name assigned for Azure SQL). This port is needed for the TDS protocol for communicating with Azure SQL like any MS SQL server.
   
2. From outside the Azure cloud environment, only ADO.net type of connections are currently allowed, so one needs to make sure to select the correct connection type in the different SQL tasks/components to be used in the Data flow task in the SSIS package.
   
3. And last but very important; this issue we hit upon when we tried to migrate heavy data. For light data i.e. for few rows it went smoothly but for lot of rows like number in the range of 2000, we were getting error saying:

[ADO NET Destination [25]] Error: An exception has occurred during data insertion, the message returned from the provider is: A transport-level error has occurred when sending the request to the server. (provider: TCP Provider, error: 0 - An existing connection was forcibly closed by the remote host.)

And later after doing a lot of hit and try and taking help from our network administrator and Microsoft representatives, we found that it was mainly because our Infosys firewall was blocking the communication during heavy data flow. Connectivity was getting dropped due to smartdefence enabled on the firewall. It was set to monitor only mode and then everything just went fine.

After last month’s release, Azure SQL now supports bulk upload using BCP along with the addition of a new Firewall support feature. As per the Azure SQL Team; “the new firewall feature allows a customer to specify a list of IP addresses that can access their SQL Azure Server instance. Security is a concern for companies looking at storing data in the cloud and with this new feature one can rest assured that only the hosts one specifies will be allowed to connect”.

This holds for any communication made from outside Azure environment (including using SSIS package on premise) and the firewall will deny all connections by default, so before using one’s own Azure SQL instance, one needs to go to the SQL Azure Portal and configure the allowed list so that existing clients can continue to connect.

Posted by Rahul Bandopadhyaya at 5:04 AM | Permalink | Comments (1) | TrackBacks (0)

Software engineering methodologies are evolving to address the complexity and probable dynamism of the business requirements. One of the methodologies –Agile - helps teams respond to the unpredictability of building software through incremental, iterative work cadences, also known as sprints. In an agile paradigm, every aspect of development — requirements, design, etc.  is continually revisited throughout the lifecycle creating high visibility of the progress to all the stakeholders in the application development process. Traditional methodology involves creating the requirements for the complete application and then the development team almost goes into hibernation to develop the application and more often than not ends up giving unpleasant surprises to business. Nobody inclusive of the SMEs sometimes is able to visualize the final application state; moreover the demands are changing and creating chaos in the development process.  AGILE comes to rescue as all the stakeholders are involved in the application development process beginning with the first iteration. Instead of building what developer understands or rather interprets and showing end product collectively, we prefer to show application in pieces take feedback and improve in an iterative manner. It becomes a continuous improvement process.  AGILE gives a good visibility on development process, application feature sets and planning process to all stakeholders.

Continue reading here

Posted by Sachin Prakash Sancheti at 1:00 PM | Permalink | Comments (0) | TrackBacks (0)

As a part of the current work we are trying to provide Federated Authentication in Azure Application. We want to develop a web application in Azure which would outsource authentication service to another component which would in turn authenticate users with its own enterprise. The basic idea is to be able to provide access to authenticated users from trusted organizations. As a sample we will use MyCompany as one of the trusted organizations. 
To achieve this we explored Windows Identity Foundation (previously called as Geneva Framework). Windows Identity Foundation provides Claims based Identity Management. It means the applications would have only authorization logic as per the claims (attributes) of the users since these claims are certified by a trusted source; these claims are secured also and hence can be trusted by the authentication mechanism. Using this framework authentication can be outsourced to some other central application or central storage and develop a claim aware application. Hence there is no need to bother about plumbing of the authentication code; one can just make use of the trusted claims received from the application which takes responsibility for authentication.

You might have got the point; this claims based model can be extended to achieve Enterprise SSO / Web SSO / Federated Authentication.

Continue reading "MYOC - Enabling Federation on Azure Applications using Windows Identity Foundation" »

Posted by Sidharth Ghag at 4:28 PM | Permalink | Comments (0) | TrackBacks (0)

Following my previous blog, on the MYOC system requirements, here I shall be covering the solution design of MYOC.

Listed below are the basic design considerations identified for the application:

1. Application must be accessible from any device and any location

2. Secure access to application resources using a federated identity and allowing users to reuse their existing credentials which they may have with other Identity providers such as Live, AD, OpenId etc.

3. Application should support multi-tenancy and have elastic scale to meet fluctuating workload demands
   

MYOC is a cloud application build on the Windows Azure platform. The above design considerations are satisfied by the Azure services platform and the .Net framework. Here is a pictorial representation of the MYOC solution:

MYOC Solution Architecture

At the core lies MYOC build on the Windows Azure platform, to provide poll functionality to a varied set of users ranging from the retail consumer to the Enterprise User. Polls can be made accessible over different devices such as PC, laptop, mobile, telephone etc.  At the same time the poll presented is context aware, with respect to the device on which the poll is being presented. Polls will be presented in a manner that is more suited to the device used by the user. For e.g., from a laptop or PC, the user will be presented a poll on SilverLight; from mobile, the Poll will be presented in plain HTML; if the user accesses the poll from a Telephone, the poll would be presented over IVR.

The earlier stated design considerations in this blog, have been addressed by identifying the following key areas:

Service Oriented Architecture

The application has been build on the principles of SOA, with access to the application functionality provided by REST based services over HTTP with the goals of having them to

1. Be easily consumed by any platform be it .NET, Java, mobile apps, widgets or any other which exists based on open standards
   

2. Be easily extendable to support additional devices for the future
   

3. Provide a development platform for other users to build poll application on MYOC
   

Having been build on top of Windows Azure, the ubiquity of this platform allows MYOC functionality to be accessible by any standard http-aware devices from any location which has internet access. Using the SOA design model achieved using REST we were able to provide users and developers with unified access on the MYOC platform across multiple delivery channels.

  MYOC Service Oriented Architecture Design

To read more on the way REST interfaces are implemented in MYOC, go here .

Federated access

MYOC has been designed to be extensible so as to support multiple identity providers such as Live, Open Id, Enterprise Ads, Custom Forms Authentication etc..The app can handle authentication and access control across these multiple identities from a single place. And this has been possible as the Authentication and Access control in MYOC has been managed separately on the cloud. This separation has been primarily achieved by using the Windows Identity Foundation (previously known as Geneva) and the .NET Service bus Access Control Service (ACS). The federation model allows users possessing separate identities to access this application in the same manner as they would experience with other apps supporting these identities. 

MYOC Access Control Design

From the above figure depicting a pictorial representation of the working model of ACS in which the user access rules is not maintained in the application code or some privately owned application repository. It will be maintained in a common repository of the Access Control Service in the form of claims. Every user (Requestor) accessing MYOC (Relying Party), once authenticated by their respective Identity providers will be provided a claim by ACS. Within the application, these claims are used by MYOC to decide the access privileges of any user. Separating the access control rules now allows architects to re-use claims across multiple applications and thus centrally managing and maintaining this repository.
Also access to the MYOC functionality is secured by the .NET Service Bus and the Windows Identity foundation. A detailed explanation on the federation process used in MYOC is available here.

SaaS Enabled

In addition to being Service oriented, a SaaS enabled application also needs to be multi-tenant and highly configurable. In MYOC, the current version, has been enabled to support multi-tenancy, leaving out configurability for a future version of the app. The various multi-tenant deployments models that can be supported by the application are represented below.

MYOC multi-tenancy models supported 

Multi-tenancy in MYOC is configured at both an application instance level and the data storage level.

Application Instance Multi-tenancy

Application instance level multi-tenancy is directly enabled and provided by the virtualized platform of Azure. Avoiding any user specific implementation in my code, I can deploy an application in either a shared model, by using the same web role instances for every user or in an isolated manner, by provisioning a new hosted instance for every customer who wishes an isolated model. 

Data storage Multi-tenancy

Multi-tenancy at the data level is what would need to be explicitly handled in the application design. The access and isolation of the customer data is configured based on the domain a particular user identity belongs. Two multi-tenancy schemas are possible in the storage aspects of the application - Shared data and Isolated data

Shared Data multi-tenancy model

In this model, the poll data created by the users will be stored in a common storage account and the data of a particular user is partitioned by their respective unique user id’s thus bringing in data isolation in a shared model. The application, based on the users identity, passes the data access request to the common storage. The application then accesses the data based on the users id which is the partition key identified,
The shared data schema is more suited for retail consumers, the likes of Live , Yahoo etc.., users, not warranting the need for any dedicated data isolation or meet any specific data security compliances. This model makes more efficient usage of the storage account as compared to the following isolated model but is perceived as less secure model to eliminate data privacy concerns

Isolated Data multi-tenancy model

In this model, the polling data will be persisted in a separate storage account and not the one used in the shared data model. Enterprises that are generally privy about their data prefer this model, wherein they get complete data isolation of the poll created by the enterprise users.
In this case the application, based on the users domain, route the data access request to a separate configured storage account. The separate account is mapped to the domain from which a user logs in. This is managed at the application level with the logic embedded in code. The model provides a more secure approach to reduce the data privacy concerns but at the same time lower storage utilization efficiency.

In the next blog I shall cover the technical architecture of the application

Posted by Sidharth Ghag at 9:47 AM | Permalink | Comments (0) | TrackBacks (0)

In this post on the MYOC series, we will look at how RESTful services can be build on Windows Azure. Building RESTful services may be no different than how you would do it traditionally on-premise. Nonethless this post will detail the steps in building REST services in Azure and additional steps which need to be considered while developing a service for Azure.

Highlighted below are the steps which differ.

REST (Representational State Transfer) is an architecture style that leverages the existing technology and protocols of the Web, like HTTP and XML. REST describes how distributed resources can be defined and addressed for simple communication and high scalability. REST specified how resources can be defined and then addressed  using Universal Resource Identifier(URI) like a web address. REST leverages http programming model and uses the same HTTP web methods for resources too –

a. GET – to get a resource
b. POST – to insert into a resource
c. PUT – to update a resource
d. DELETE – to delete a resource

REST services are easily discoverable through Web URLs and can be consumed by any http client, thus offering a great deal of interoperability between heterogeneous systems. 

WCF (Windows Communication Foundation) is Microsoft’s unified programming model for building service-oriented applications. It enables developers to build secure, reliable, transacted solutions that integrate across platforms and interoperate with existing investments.  A WCF Service uses three types of contracts –

a. Service contract to communicate with its client
b. operation contract for each of the methods in a service
c. Data contract  for request and response data

In WCF Service, the REST based call of the service methods will be allowed by specifying WebGet (for GET method) or WebInvoke(for POST method) attribute in the service interface.

Following are the steps to create REST based WCF service using Visual Studio 2008 –

1. Open the Visual Studio 2008 and create a cloud service project. Right click on the web role project and click on Add -> New Item. All the aspx pages, web services and WCF services are required to add in web role project. [Additional step for Azure Projects]

2. Now select WCF Service from the templates and provide a name of service and click on Add button.

3. This will add 3 files in the project –

a. IMyTestService.cs – the interface file which contains method declarations.
b. MyTestService.svc – the mark up file which contains information about service like code language, mode of application, service namespace, name of code behind file etc.
c. MyTestService.svc.cs – the code behind file which contains actual implementation of the methods declared in the interface definition.

4. The <system.serviceModel> section of the web.config of the project should be as shown below –

<system.serviceModel>
        <behaviors>
            <serviceBehaviors>
                <behavior name="MyCloudApp_WebRole.MyTestServiceBehavior">
                    <serviceMetadata httpGetEnabled="true" />
                    <serviceDebug includeExceptionDetailInFaults="false" />
                </behavior>
            </serviceBehaviors>
            <endpointBehaviors>
              <behavior name="webHttp">
                <webHttp />
              </behavior>
            </endpointBehaviors>
        </behaviors>
        <bindings>
            <webHttpBinding>
                <binding name="webHttp" maxReceivedMessageSize="50000000" maxBufferPoolSize="50000000">
                    <readerQuotas maxArrayLength="50000000" />
                </binding>
            </webHttpBinding>
        </bindings>
          <services>
            <service behaviorConfiguration="MyCloudApp_WebRole.MyTestServiceBehavior" name="MyCloudApp_WebRole.MyTestService">
                <endpoint address="" binding="webHttpBinding" contract="MyCloudApp_WebRole.IMyTestService" bindingConfiguration="webHttp" behaviorConfiguration="webHttp" />                   
                <endpoint address="mex" binding="mexHttpBinding" contract="IMetadataExchange" />
            </service>
        </services>
</system.serviceModel>

5. Create a Data Entity class file with Data Contract attribute as mentioned below –

a. Add reference of ‘System.Runtime.Serialization’ and add below namespace after existing ones

using System.Runtime.Serialization; 

b. Decorate the class with DataContract attribute and class members with DataMember attribute

namespace MyCloudApp_WebRole
{
    [DataContract]
    public class Poll
    {
        [DataMember]
        int pollId;

        [DataMember]
        string pollName;

        [DataMember]
        string pollQuestion;
    }
}

c. DataContract attribute has following input parameters –

i. IsReference – boolean value to indicate if data contract is a reference
ii. Name – name of the data contract for the type
iii. Namespace – namespace of the data contract for the type

d. DataMember attribute has following input parameters –

i. EmitDefaultValue – boolean value to indicate if there is a default value
ii. IsRequired – boolean value to indicate if the data member is mandatory field
iii. Name – name of data member
iv. Order – order of data member to appear in the data contract

6. Now open the IMyTestService.cs file and perform below steps –

a. Add reference of ‘System.ServiceModel.Web’ and add below namespaces after existing ones

using System.ServiceModel.Web; 

b. Add method declaration in the service interface and decorate with WebGet / WebInvoke attribute

namespace MyCloudApp_WebRole
{   
    [ServiceContract]
    public interface IMyTestService
    {
  [WebInvoke(Method = "POST", UriTemplate = "Poll")]
  [OperationContract]       
  void CreatePoll(MyCloudApp_WebRole.Poll poll);
   
        [WebGet(UriTemplate = "Poll/Id/{pollId}",
        ResponseFormat = WebMessageFormat.Xml,
        BodyStyle = WebMessageBodyStyle.Bare)]
        [OperationContract]
        MyCloudApp_WebRole.Poll GetPollbyId(string pollId);
    }
}

c. Follow the guidelines to create a URI template for a service method.
 
d. ServiceContract attribute is used to define a service contract. Service contract is exposed to the service consumer to call the service. Service contract attribute has following properties –

i. CallbackContract – specifies the return contract in a duplex conversation
ii. ConfigurationName – the name of the service element in the configuration file to use
iii. HasProtectionLevel – indicates whether all messages supporting the contract have an explicit ProtectionLevel value
iv. Name – name of the contract in the WSDL <portType> element
v. Namespace – namespace of the contract in the WSDL <portType> element
vi. ProtectionLevel – the level of protection for the messages
vii. SessionMode – specifies whether the contract requires a binding that supports sessions

e. OperationContract attribute marks the methods exposed by the service in the service contract in a WCF service. it has following properties –

i. Action - specifies the action performed by the operation.
ii. AsyncPattern - indicates that the operation is implemented or can be called asynchronously using a Begin/End method pair.
iii. HasProtectionLevel - indicates whether the ProtectionLevel property has been explicitly set.
iv. IsOneWay - indicates that the operation only accepts input message and doesn’t return any output message.
v. IsInitiating - specifies whether this operation can be the initial operation in a session.
vi. IsTerminating - specifies whether WCF attempts to terminate the current session after the operation completes.
vii. ProtectionLevel - specifies the message-level security that an operation requires at run time.
viii. ReplyAction - specifies the action of the reply message for the operation.

f. WebGet attribute indicates that a method can retrieve values from Web. WebGet is used for the operations having HTTP GET verb. It has following properties –

i. BodyStyle – format of message body; wrapped or bare
ii. RequestFormat – the format of input message; JSON or XML
iii. ResponseFormat – the format of output message; JSON or XML
iv. UriTemplate – the URI for the resource to be accessed

g. WebInvoke attribute indicates that a method can be invoked by Web. Web invoke is used for the operations having HTTP POST, PUT, or DELETE verbs. It has following properties –

i. BodyStyle – format of message body; wrapped or bare
ii. Method – the HTTP verb used in the operation; POST, PUT, GET, DELETE
iii. RequestFormat – the format of input message; JSON or XML
iv. ResponseFormat – the format of output message; JSON or XML
v. UriTemplate – the URI for the resource to be accessed

7. The MytestService.svc file is a markup file which will contain the information about the service which will look like –

<%@ ServiceHost Language="C#" Debug="true" Service="MyCloudApp_WebRole.MyTestService" CodeBehind="MyTestService.svc.cs" %>  

8. The MyTestService.svc.cs file will have the implementation of the methods defined in the interface class IMyTestService.cs which will look like –

namespace MyCloudApp_WebRole
{
    // NOTE: If you change the class name "MyTestService" here, you must also update the reference to "MyTestService" in Web.config.
    public class MyTestService : IMyTestService
    {
        void CreatePoll(MyCloudApp_WebRole.Poll poll)
        {
            // write code to create poll
        }

        MyCloudApp_WebRole.Poll GetPollbyId(string pollId)
        {
            // write code to Get poll details based on input poll ID
        }
    }
}

Even consuming a service on Azure is almost same as consuming a service in on-premises application. We’ll have a look on how the REST based WCF Service can be consumed by a client asynchronously using  WebClient –

1. Write the following code when CreatePoll service method is to be called –

// take user input for poll creation and save it in Poll entity object

// since data contract cannot be exposed to client directly, create a string called xmlString by serializing Poll entity object

// create service URI and web client
string strUri;
if (current.Request.UserHostName == "127.0.0.1")
{
    // URI for running the project on dev fabric
    strUri = "http://" + current.Request.Url.DnsSafeHost + ":" + Request.Url.Port.ToString() + "/MyTestService.svc/Poll";
}
else
{
    // URI for hosted environment
    strUri = "http://" + current.Request.Url.DnsSafeHost + "/MyTestService.svc/Poll";
}            
           
Uri uri = new Uri(strUri);
WebClient client = new WebClient();
client.Headers["Content-Type"] = "text/plain";

The highlighted text above is for us to be able to figure out whether the code is being executed from the dev fabric or from the Azure environment. [Additional step for Azure Projects]

2. Call the service asynchronously using UploadStringAsync to POST the data to create the poll to the service

// send the string to specified resource asynchronously 
client.UploadStringAsync(uri, xmlString);  
// Create the handler to be called when upload is completed 
client.UploadStringCompleted += new UploadStringCompletedEventHandler(PollUploadCompleted); 

3. Create the handler and write code to be called after upload completion

void PollUploadCompleted(object sender, UploadStringCompletedEventArgs e)
{
      // write code to perform operations after upload completion like poll creaion message to user etc.
}

4. To call GetPollbyId service method with GET operation, follow the step 1 mentioned above with modified URI for the method

// get the pollID for which poll details are to be fetched

// create service URI and web client
string strUri;
if (current.Request.UserHostName == "127.0.0.1")
{
    // URI for running the project on dev fabric
    strUri = "http://" + current.Request.Url.DnsSafeHost + ":" + Request.Url.Port.ToString() + "/MyTestService.svc/Poll/Id/" + pollId;
}
else
{
    // URI for hosted environment
    strUri = "http://" + current.Request.Url.DnsSafeHost + "/MyTestService.svc/Poll/Id/" + pollId;
}            
           
Uri uri = new Uri(strUri);
WebClient client = new WebClient();

[Additional step for Azure Projects]

5. Call the service asynchronously using DownloadStringAsync to GET the poll details for a given poll ID

// download string from the specified URI  
client.DownloadStringAsync (uri);  
// Create the handler to be called when download is completed  
client.DownloadStringCompleted += new DownloadStringCompletedEventHandler(PollDownloadCompleted);   
  

6. Create the handler and write code to be called after download completion

void PollDownloadCompleted(object sender, DownloadStringCompletedEventArgs e)
{
      // write code to perform operations after download completion like displaying poll details.
}

After following the steps above, we can conclude that creating a RESTful WCF service on Azure is almost same as creating an on-premise RESTful WCF service.

Posted by Sidharth Ghag at 3:27 AM | Permalink | Comments (0) | TrackBacks (0)

Life is never too simple! And who can know this better than consultants. Consultants have their own reasons to follow an approach for achieving a business need using any COTS solutions. However, in my personal experience I feel that any packages product should be used with utmost care and be modified after following some standard principles. There would be lot of pressure from some members of the client team to customize the product so that it has look and feel of the previous application and which lead to reduction of training needs. On the other hand, the IT teams would like to keep it simple to reduce the upgrade costs and the pains associated with it along with recurring cost of maintenance. The aim is to reduce the overall TCO (Total Cost of Ownership) and the APM(Associated Pain of Management). Well the second acronym is my brainchild.

Of course, there are several factors which do affect the reason to choose the path of customizations. I had some thoughts around this which is available in one of my older blogs entry The “framework” conundrum in ERP/CRM

If we keep these factors on the side; I think there are some basic principles which should be discussed with the client beforehand and whetted before the solution development can happen. This is important to set the broad guidelines for the long match which is going to follow between consultants, business users, IT users and the CXO community at large. Once the base is defined; I have found it easier to discuss the greater details around the solution-design with the stakeholders. Any aberration from the basic agreed upon principals can be “striked-off” by simply referring back to the ‘Commandment’ being violated!

Here is the list of ‘Commandment’ that I personally follow to start with. Depending on client situation this can get modified; but I can assure you that it works well in most of the cases with the clients sometimes coming back to them after some initial modifications when they understand the ramification of not following them in spirit.

So, here goes my personal list (which I call TRY TO GRASP) –

1. TRY - The product should be used out of box as much as possible.
2. RECOGNIZE - The entity structure and relationships as defined in the product should be modified to the minimum.
3. YEARN - ‘Discuss’, ‘Seek Advise’ & ‘Search’ for workaround without compromising on point 1 & 2.
4. TOTAL VIEW - Look at the overall process flow (as needed by Client’s Business) in ERP before finalizing the elements / entities. Do not think at ‘module’ level!
5. OBSERVE - Just using an entity (because it sounds similar to client needs) is not the right way; there can be hidden issues due to product data model.
6. GRIP - Do not try and compromise on the ‘Financial Structure’ in the product; the solution should not interfere ‘with’ it and should be ‘around’ it
7. RESEARCH - Look for ‘certified’ solutions available from partners in areas which are gap in the product.
8. ADAPT - If nothing work – ‘Customize’. Customizations are a ‘necessary evil’.
9. SUSPECTS – Be aware of the usual suspects. Never count the number when seeing customizations; it’s the dimensions of ‘complexity’, ‘risk’ & ‘impact’ that count.
10. PEEP - Do not jump on to customizing ‘white spaces’; it could be in the roadmap of the product!

I am sure there are many other things that we follow without formally writing them down in a structure. Please share your thoughts. Are there any additional “Commandments” that you follow?

Posted by Sachin Bery at 11:00 AM | Permalink | Comments (0) | TrackBacks (0)

In my previous blog post, I had shown you a deployment model in which an application could leverage the capabilities of both on-premise as well as being on cloud. In that blog I had shown how application storage had been migrated from on-premise to the cloud.
In this post I will show you yet another deployment model possible on Azure and which may interest most of you who may have applications running on non-Microsoft technologies. I will discuss a model on how applications build on Java technologies can be deployed on Windows Azure and hence can reap the benefits of Cloud Computing.

More on "Infosys mConnect" in a Microsoft Paper is available here

With the November release of the Azure SDK, Microsoft has opened up the Windows Azure platform to applications build on non-Microsoft technologies such as J2EE. Azure can now host any application which runs on open source platforms such as Apache Tomcat, MYSql etc. In this Blog post I will walk you through one such scenario where we have migrated one of our existing non-microsoft application on to the Cloud.

Background:

Infosys mConnect is a “context aware” & “device agnostic” platform that helps to mobile enable any existing application or create one from scratch. Infosys offers mConnect, as  a product that enables Web sites, e-commerce, and banking platforms to support mobile devices without costly modifications to their services. The core differentiation of the platform lies in the way it extends traditional application functionality to any device such as a Palm, Blackberry, iPhone etc... in a way that is optimized for that end user device and the network.

Challenges:

With mobile becoming one of the de-facto delivery channels in reaching out to customers or partners for most Enterprises today; scalability and performance are becoming the key challenges faced. With the wide scale penetration of mobile devices, B2C scenarios are seeing huge demands from the growing mobile user community in terms of being able to deliver rich content in real time. Infrastructure build for catering to mobility scenarios will have to stand up and be resilient to meet the growing load and scalability demands of the mobile community. And building such an infrastructure will require significant investments to be made.
While  Infosys mConnect being a mobility platform addressing issues like device diversity and optimizing application response as per different mobile devices but with porting of mConnect applications to Azure, we could very aptly address the scalability and performance concerns. In the traditional model, mconnect was deployed within Client locations or sometime within our own Data centers. The challenges here had been with having to over-provision and at the same time manage these systems thus incurring significant upfront capital investments. With mConnect on Azure we are now able to have an offering to deal with the huge appetite of the B2C mobility service for our clients and that too with low upfront investments

Solution

Infosys mConnect is a gateway which sits between the end-user & the enterprise. The gateway, services requests from the end user which it can receive from any of the different mobile devices out there today. The request received is forwarded to the enterprise systems in the backend to retrieve the required information. Information received from the enterprise systems is then encoded in the relevant protocols into screens supported by the mobile device. The gateway based on the device info is able to gather the device context and renders screens which are suitable to the device.

mConnect Architecture 

Infosys mConnect has been built on J2EE technologies and hosted on Apache Tomcat. Also the application can be configured to work with any relational databases data storage. 
With the latest Nov SDK support for “External endpoints for worker roles and access to role configurations“ is made available on Azure. Using these features we were able to move the mConnect technology stack onto Azure AS-IS and that too in the least invasive way thus saving us considerable time and effort in the process. We were able to deploy the entire application on to Azure in 3 weeks.

mConnect deployed on Azure 

The mConnect application was deployed to Azure with SQL Azure as the Database. The application start up is handled by the Azure worker role. Before strarting up a Tomcat instance, the worker role will first map the tomcat listening port to the worker roles endpoint such that all request arriving at the worker role endpoint will be directed to Tomcat. Once the mapping has been done the worker role will fire up a tomcat instance which will inturn initialize the mConnect application.

The implementation details of running a Java app on Azure can be easily understood by downloading the Infosys developed, Tomcat solution accelerator on Azure which is available here

Benefits of Moving to Windows Azure:

1. The instant benefits we got from Moving to Azure was ability to gain instant scale without investing a single dime on Infrastructure
2. Also by using SQL Azure as the data storage we were instantaneously able to leverage relational data access capabilities on cloud with changing a single line of code
3. By adopting to a least-invasive approach to application migration, we were able to migrate a Java application on to Windows Azure in a matter of weeks and thus reducing the overall time to market

Posted by Sidharth Ghag at 10:54 AM | Permalink | Comments (0) | TrackBacks (0)

Posted by Saju Sankarankutty at 3:42 AM | Permalink | Comments (2) | TrackBacks (0)

In this blog, I will continue the discussion on a possible migration scenario of migrating existing ISV application to Azure. If you haven’t read my previous blog on this topic, then you can go here...

Ubik.com Application Architecture

Ubik.com(Presentation)

With Ubik.com users are able to create personalized mobile enabled web sites. The sites created by the users primarily contain information on the site structure and content both in textual format as well as binary images. The site structure and definitions of the sites created are contained in a standard publishing format known as Xdime. The site definitions and content along with the images selected by the users are published to the newly build web services.

Ubik.com Services:

The Ubik.com services are RESTful web services primarily responsible for managing the Xdime content and Images which are stored in the Azure Blob Storage. The web services run on the Azure operating system and communicate with the Ubik.com application over http.  The REST services provide interfaces to not only publish the contents and images into the Blob storage but also enable accessing the site content directly over the mobile devices.  Let look into the design of these service interfaces and with them being RESTful
There are two distinct set of services created:
1. Manage XDIME content
2. Manage Images for site
Each of these services support standard http GET, POST, PUT and Delete operations. 
1. The GET operation is used to create new site content or images into the Ubik Blob storage.
2. POST/PUT operations can be used interchangeably to update content/images into the Ubik Blob Storage.
3. DELETE operation is used to delete the content/images from Ubik Blob storage. 
Additionally the application also supported the capability to store data in different Blob containers based on the deployment region that the application operated in. Like any product release management, an application could be in different stages of a release say test, staging etc.., undergoing several round of tests against different versions of data sources and by different user groups (say business, devs, testers etc..) before being finally “Released to Market”. Ubik.com had a similar requirement wherein the application needed to be tested across several regions before it is released. And it was required that a single instance of the new service running be able to handle data for the different regions without having to host separate Azure instances for each region.
Based on the service URL “environment” value (env) ,as shown in the URL below, the services were able to dynamically identify  and store data in separate regions.

A standard service URL to call UBIK.com services would look like
http://[Host]/[doc|img]/[env]/[docname|imagename]

Note:  The URL’s shown here are only for representative purposes to explain my point and are not the live ones used in the application
Where ,
• Host:  DNS host name
• [doc/img]:
o doc: an url path for indicating that a content file is being handled
o img: an url path for indicating that a image is being handled
• [env] : Is the environment from which the data is being managed eg: dev , test, prod.
• [docname|imagename]: is the name of the content which is being handled from blob storage
As you can observe, the URLs are pretty intuitive and using the appropriate http URL’s with the corresponding http verbs will provide access to the required information from the Ubik storage
The error notification from these services are communicated to the client are over standard HTTP response formats containing the standard response codes such as 400,404, 500 etc, and status descriptions.

Data Storage:

The site content and images are persisted in the Azure Blob Storage. In the initial phase, primarily two separate containers were created to persist Ubik.com data, one container for site content (Xdime) and the other for storing images.
From a security aspect, the BLOB storage containers are marked as “Private” and access to the content is restricted. The Blob access is available only through the services by using the Authorization key which is available from the Azure portal as a part your storage account.
In this way I have demonstrated how we have helped one of our ISV customers adopt Azure in their application scenario. In case you have more interesting scenarios you are working on or have seen and would like to share, then do post in your comments.

Posted by Sidharth Ghag at 12:22 PM | Permalink | Comments (0) | TrackBacks (0)

Infosys on its part is helping Enterprises and ISV’s to adopt cloud.

Adopting this new paradigm of cloud computing; newer and innovative styles of using the cloud platforms will have to be explored. Here I shall walk you through one such case which demonstrates how we’ve helped one of our ISV customers, Volantis ,  to adopt cloud.  The detailed case study on this project done by Infosys is available here

Volantis is a developer of innovative solutions for mobile carriers. The company’s Ubik.com software makes it simple for users to point and click to create custom Web sites that are optimized for mobile devices.Ubik.com is a free online service that allows small businesses and consumers to quickly build a mobile Internet site, without having to write a single line of code.

To cater the applications non-functional requirements of being highly available and achieve global class scalability, which would help meet the demands a rapidly growing user base, Infosys helped Volantis to offload Ubik’s data storage on to the Microsoft Azure platform.

Background
Volantis was looking at migrating some of their existing applications functionality to the cloud and gain the benefits of a cloud deployment to meet the high demanding non-functional requirements of the application with near-zero investments. Infosys helped Volantis identify the scenario that could demonstrate the benefits of Windows Azure cloud. The scenario identified was in addressing the scalability demands of the storage requirements in the Ubik.com application.
 

Challenges & Goals:
The main challenge faced by Volantis was to store large amounts of user created site content & its associated metadata, being as large as over 30 terabytes. And that too without having to actually invest in setting up such an infrastructure upfront from their own working capital.  Volantis desired that the Ubik.com storage be made capable to scale seamlessly and meet the storage demands of  their mobile users community.

Proposed Solution:
In the process Infosys proposed a phased approach for migrating Ubik.com to Azure. In the first phase, and which I would be touching upon here, the migration of the file storage to Azure BLOB storage was proposed as it was a layer with minimal change and the risk associated with the change was low. With this approach the benefits of the cloud could be easily realized in a short span of time.
Our solution proposed architecting a separate RESTful services layer on top of the Blob Storage that would provide a seamless and scalable access to the Ubik.com application. Exposing RESTful service wrappers on the Azure operating system was essential so as to have blob storage accessible from Ubik’s non-microsoft application APIs.
A separate service layer was build to minimize the changes in using Azure Storage with Ubik.com. Site content uploaded by the Ubik.com users into storage comprised of images and XDIME files. These files before uploading into the storage had to be parsed and then appropriately persisted. The newly created services layer on Azure handled this and thus avoiding making any significant changes to the existing application codebase.

Migration Approach:
The migration of Ubik’s data storage from on-premise to the cloud was done as shown in the figure below:

Team Dynamics:
Both the Infosys and Volantis developers worked as a team to identify areas in code which could have a possible impact owing to the change. The team also identified service interfaces which would be required to be exposed from the Windows Azure end to make the storage directly accessible on the Ubik’s non-Microsoft platform. The requirements and design was done collaboratively between the Infosys & Volantis teams working across different timezones. Using basecamp as the project management and collaboration tool helped to better cordinate the project across different time zones and deliver the project in a short span of time.
Once this initial design completed, both the teams got about working on their respective areas of the application. The Volantis team made changes in the Ubik application code and the Infosys team focused on building the RESTful services on the Azure operating system.  On code completion, the teams got together to test the services and the integration touch points. The entire project was completed within four weeks. 
Here I have shown a typical working model by which we engage with customers to help adopt cloud. With Volantis, Infosys not only assisted in identifying a cloud scenario but also helped in migrate an existing application to the Azure cloud.
Continuing on my Volantis experience, in my next blog, I would describe the technical architecture of the Ubik.com application on Azure.

Posted by Sidharth Ghag at 7:35 PM | Permalink | Comments (0) | TrackBacks (0)

MOSS 2007 framework came with enough built in capabilities that could be exploited to make a SharePoint site look and feel unlike the typical out-of-the-box SharePoint site. Unfortunately, too often, developers either lacked the training or exposure to simple techniques that could be used to ‘brand’ their sites or there was not enough business justification to spend time working on that.

With upcoming SharePoint 2010 and related suite of products, the focus is going to expand from intranet and collaboration oriented internal sites to developing external facing web artifacts.  This was clearly highlighted by Steve Ballmer himself during the launch of SharePoint 2010 earlier last month. This will require higher level of attention being paid to branding aspects of SharePoint.

Over the next several weeks, I plan to share my insights into how SharePoint 2010 facilitates easier branding and building custom themes or skins. Idea is to understand how users can still tap into every built-in capability of this powerful platform without compromising on the design and branding aspects. Branding imperatives like establishing corporate identity and ownership, reinforcing the enterprise standards and creating a sense of place/connection with intangible brand values are key in making external facing artifacts successful and we will see how the new platform supports that

For this post, I want to share couple of thoughts about an ideal ‘end state’ I would like to see products like SharePoint reach when it comes to supporting branding work.

Posted by Chandan Gokhale at 5:20 PM | Permalink | Comments (0) | TrackBacks (0)

Today, Atul Gupta and self delivered a session on Architecting for NFRs at Microsoft Virtual Tech Days event.

Non Functional Requirements (NFRs) is one of the very important system attribute and help in defining the Quality of Service (QoS) for the System. Examples of QoS are Availability, Performance, Scalability, Reliability, Usability, Extensibility, etc. These need to be planned right from the project initiation stages.

We talked at length on this at MS Virtual Tech Days; the presentation makes an attempt to provide the guidelines while architecting the applications for meeting these QoS parameters and can be downloaded from here.

Posted by Sudhanshu Hate at 9:32 AM | Permalink | Comments (2) | TrackBacks (0)

Posted by Rajesh Gupta at 3:08 AM | Permalink | Comments (0) | TrackBacks (0)

How would Cloud Computing help an Enterprise bring about IT consolidation, an approach that can help reduce ongoing running costs as well as help respond to business needs faster?

Enterprises globally are growing at a rapid pace both organically and inorganically. A few years back IT build systems for businesses which would last a time frame ranging anywhere from 5 to 10 years. Systems used to be built to last. But today, systems have to be designed to handle change. Change, which is inevitable today, is not in years but months and sometimes even days. The IT delivery cycles are expected to shrink and meet the demands of the business faster.
Click here to continue reading..

Posted by Sidharth Ghag at 5:37 PM | Permalink | Comments (0) | TrackBacks (0)

MYOC-Make Your Opinion Count

Objectives

MYOC is conceptualized as a polling platform where consumers can use it in a more social community setup to seek opinions from family, friends etc. Similarly enterprises can use this platform to seek opinions from key stakeholders, be it employee, suppliers or customers. MYOC is built as a platform and not an application. Unlike an application the platform will offer other developers the opportunity to consume MYOC services and provide enriched functionality of their own thus expanding the reach of the application.

Use cases:

Below is a pictorial representation of the applications navigation flow

Technical challenges:

Some key areas which we will be required to address from a techology perspective are:

a. Ability to participate in a poll from any location and using any device 
b. Users must be able to access the application using their own identities. Nultiple Identity providers such as Live, Enterprise Active directory, Forms Authentication, Open Id etc.. should be supported and be configurable to support new Identity providers required in the future
c. Application should be SaaS ready with the capability to be multi-tenant and scalable to meet fluctuating Internet demands
d. Appication should have a service platform where application functionality can be offered as a service to support future customizations and introducing new devices or delivery channels easily such as widgets, mashup apps on facebook,  etc..

As you can see from the use case and some of the challenges listed above, these are standard requirements for any public facing web application sans the cloud. The question then is, What makes MYOC a cloud app? The difference here lies underneath the facade of the application, which is not very evident from the outside, and where lies the true value of the technology. From outside it may seem a regular web application having a regular public facing URL but the construction of MYOC and the technology underpinnings is where the differences would lie. I would be covering this aspect of cloud development in my forthcoming blogs.

In my next blog, I will be sharing the application solution design.

BTW, You can try out the application here. If you any feedback, please do share.

Posted by Sidharth Ghag at 1:54 PM | Permalink | Comments (0) | TrackBacks (0)

A little later while doing a Silverlight 2 training, I realized that I would have to go back and forth between my slide deck and visual studio and expression blend and to me that seemed non-productive. Hence I ended up building the entire presentation as a Silverlight application and I integrated all demos in the same application. So it was all the time IE in which I ran the application. I had even added a slight animation in a corner to run for 1 hour and that gave me and the audience a good idea of elapsed time and the time remaining. Obviously I had to ensure that I ended on time for this to look good.

Needless to say that when recently I started working with Sketchflow, it made sense to do the presentation using content built in Sketchflow itself and that’s what I did. An interesting learning on visual state manager (that I had earlier written about here) was the fact that states in a state group are mutually exclusive, and that the application/control can exist in multiple states at the same time if they aren’t in the same group.

For the presentation I was making, I had a scenario where I was animating some content and when it would stop I had to trigger animation of some more content. This isn’t possibly on it own with Sketchflow animation as the animation will play from start to end and any new animation will start from the base state. The same is true with states in a state group. Finally I realized that I could actually create many different state groups and add my required animations as states inside of different groups. Now when any new state animation will not start from base state, but from the state the application/control was in, at the end of previous state animation.

Posted by Atul Gupta at 10:22 AM | Permalink | Comments (0) | TrackBacks (0)