Amazon for Cloud Computing is as Starbucks for Coffee??
I’ve been globe trotting lately - India, UK, US and such. Living in Seattle puts a tacit yet obliging pressure on you to visit Starbucks. Ordering a cup of coffee after 9 months was an all new experience from the previous time I was there. Double-tall, non-fat, de-cafe, extra-hot cappuccino for instance, Sugar free, soy, cinnamon dolce, and no-whip latte for another. So I couldn’t stop but notice the granularity of the orders and what Starbucks had accomplished in the past 9 months.
A year or so ago Starbucks introduced “our promise” concept which promised the customers that they will brew the drinks exactly how they want it. It involved soy milk, brown sugar and a few other options. A slight dissatisfaction would get you a whole new cup of the same drink but with corrections. They did not hesitate until you were satisfied. Once I requested my tall cappuccino be heated and they gave me a new extra-hot cup of coffee – just the way I liked it. I paid $2.85 for the drink and left the store with a sense of satisfaction.
But now, it is a whole new world out there. Customers seemed happier getting their drinks. The more complicated the orders were the broader the smiles were. Bingo!!! That is what we need in Cloud Computing too! I will come to this in the next paragraph. I ordered tall cappuccino but I was recommended that I get a double tall, extra-hot, wet cappuccino. I paid $3.41 but I left the store this time with a sense of delight.
Alright! We have established I like cappuccino. But we can’t miss but notice what Starbucks has achieved with coffee. They have brought in a host of syrups, added a twist of health and wrapped it with a personal touch of “promise” = customized drinks = perfect mornings = happier customers.
Can Amazon and Google achieve this with cloud computing? Can we have “double RAC node, four WebLogic instance, JMS messaging enabled Sterling DOM supported on RAID 5 disks” or a “DB2 based, single instance WebSphere application server hosting Manhattan forecasting engine”?
In my opinion we as service providers should take the onus of encouraging the SCM package vendors to come-up with a “Cloud License” model which takes machine instances to consideration. Dynamically adding a machine instance on-demand may seem hunky-dory but preparing a new machine instance to have the same execution environment as the one that crashed isn’t easy for multi-tiered applications. Besides, instantiating another virtual machine may only work for simpler standalone applications.
Multi-tiered SCM applications require a host of prerequisite installations and configurations to fall in-place before you can even attempt to start a basic version of the engines. Therefore, we need to come-up with cloud licenses which can be maintained as a repository on the cloud. Creating the execution environment is then a matter of following step-by-step instructions that can perhaps be automated.
A host of such vendors will help widen the repository and help Cloud Infrastructure providers like Amazon and Google brew different production environments and attract variety of user communities providing environments exactly how they’d like it.
Hurry up guys IBM is already marking its space. The sooner the better! No?