Why should you be in the cloud? The question is why aren’t you there yet? Let’s take a look at traditional IT with a business-to-consumer retail sales business. Let’s say your store is running on several horizontal nodes for the application server. You also need a load balancer, and caching proxy. Additionally you have several HTTP servers serving static data. Finally, you have a database for all of these application servers to talk. On top of this you need a disaster recovery infrastructure.
Now all this is running fine and your store is doing great. Looking toward the future, you start planning for your peak demand season. This could be numerous products: if you sell flowers, graduation, Mother’s Day, and Valentine’s Day will be big for you; if you sell general retail items, from Black Friday until Christmas Eve will be huge for you. What does this planning look like?
Under the traditional mode of planning, it might go something like this. First you would want to review last year’s volume, see what your current capacity is, and how much additional volume you think you can support. If, from this analysis, you need more capacity, that will lead to the following tasks. Procure more systems and have these added to the data center. Then, ensure you have the correct number of licenses for the software you need in the stack. Now, load the operating system (OS) and other security and performance settings the operating systems need. Next, install the software onto these new systems. Finally, you need to integrate these into your existing cluster, test the integration, and ensure all the approriate performance settings are applied to these new app servers.
Whoa! I am tired just thinking about all that, and your IT staff probaly starts to look like the guy in the picture. That certainly does not sound like we are making optimal use of time, hardware, or software. What do you do with that hardare after the peak period is over? Do you decommission it and move to another project? Which would most likely cause you to reimage the OS, and possibly even move it, and now deal with who pays for the hardware? This certainly is not a green computing.
So what is the solution? How do we avoid this endless churn getting ready for peak periods? What if you are like that flower seller who has several peak periods during the year—do you need to keep that hardware idle during most of the year just to avoid the trouble of removing and adding from your cluster all the time?
This is where the cloud comes in.
I am not trying to say this is the be-all-end-all answer, or is your knight in shining armor.
But, it can solve a lot of the problems we discussed. No matter if you are in the public or private cloud, cloud can help address a lot of these issues more easily.
Look at this same scenario now that you are running your store in the cloud.
A peak season is coming, you analyze the capacity that you need, and realize what you have is not enough. So you decide to spin up (create and start) some new images. These images are based on your specified image, so no time is spent getting new hardware and ensuring all items are installed correctly, and performance tuning settings are applied and other apps are installed. This is all taken care of for you by the provisioning software and you now have your capacity. What would have taken weeks before or even months, now took at most two days to get these up and running.
