In my previous article, I discussed several different points about moving to the cloud. A lot of companies do it for many reasons, but the majority of them in my experience normally come down to two major reasons: 1) Financial 2) Security of business. The security tab isn’t the same as say a password or encryption; I mean it as a form of reassurance in the case of disaster or redundancy.
Lets take the first reason in this article: Financial. Many people incorrectly assume that moving everything to the cloud can save them millions of dollars, and they can see an immediate return on their investment right away. Bad news people, that’s not normally how it goes.
Remember, that no commercial hosting cloud provider does it for free. And if you have your own cloud stood up at a hosted colocation, you are still paying fees for the three “Ps” (ping, power and pipe). The next expense that does not go away quickly is people. It takes knowledge to migrate your systems and /or applications to the cloud. Even if you fire your entire staff, and had a third party come in to do the work, you will still spend more money on that happening, because they have to learn your systems before they can accomplish your migration.
Alternately, if you are planning your cloud migration, and you are going to use your own internal resources to do it, the big question is: “Are they ready?” Going back up to the two reasons why companies migrate to the cloud, Financial and / or Security of business, this question encompasses both. How?
Lets look at the Financial side this month. You will need to spend money on standing up an in house test cloud. Many people already have virtualization using many types of Hypervisors, but cloud is different. Cloud systems use hypervisors as a means to an end. An example is some hypervisors you pay for, while others are open source. The test cloud can be completely open source, or you can use your favorite vendor’s software. Most software costs money, so try as much open source stuff as possible for testing and learning functionality. The main point is, get your engineers up to speed with using cloud. Its costs money to get them up to speed (hourly wage and such), but you will need them in the future.
Lets look at it more closely now. Lets say you use an open source cloud tool. You have to have systems to run that on, even if they may be virtual (costs money). Then you need to make sure that your network people get engaged, because the cloud will want to either setup its own network, or it will need to be tied to your existing one to hand out IP addresses to the various cloud instances (costs money). Your network engineers need to make sure that you are not infringing on the production network in any way. Alternately, if you are running a test cloud from a hosted provider (costs money), they will need to make sure that the data from the provider does not have precedence over your exist network traffic.
The next people that you need to engage are going to be your information security people (costs money). Setting up the test cloud will require strong passwords on user accounts, certificates if possible (even if they are self signed), and node protections (hardening, antivirus/malware and log interrogation). Even though this is test, you need to start making the needed preparations as if it wasn’t. This could be a machine image(s) you already use, or maybe you need to create one that will support the cloud software (costs money).
Finally, now that you have a test cloud setup for your people to test out and learn with, you need to get one other group of people involved; your developers (costs money). One thing that many companies make mistakes on, is not bringing your developers in soon enough. These people are what make your applications and financial systems work for you on a software level. Have them test their code in your test cloud. Not just the final code they copied out of production, but each step of the development lifecycle: Proof of Concept, Development, System Test, Integration Test, User Acceptance, Staging and Production. Some test clouds will allow the developers to spin these environments up on their own, with operations assisting them. Have your developers test everything, just like the infrastructure engineers.
Bottom line, I want readers to understand that it costs money, even if their goal is to save money in the long run. Regardless of your ROI that you have built, you will be spending money getting ready for your cloud migration. As long as you understand that, and budget for it, you should be fine.
My next article will cover the other aspect of the Planning part, the Security of business.
By Richard Thayer
