Real life moving tech from On-premise to Cloud

For a lot of the major companies, the quick wins of moving to the cloud, which really translates to making it easy to vertically scale some existing servers is over.

We are now getting into the area of integrating serious existing systems with cloud services and having to match the expectations of cloud services to everything else that exists in the company.

One of the big things I’ve seen now multiple times is that we are trying to match the “everything is instant” “everything is real-time” expectations of consumer level products that are built totally in the cloud with existing on premise infrastructure and platforms. “What is the problem?” I hear you cry, just upgrade everything, but it doesn’t quite work like that. Now you have a cloud system that has an expectation of receiving data from internal systems, and indeed can pull data from said systems at real time, but the source systems just can’t take that load, everything from database servers, to internal network, to app servers are not designed for such a load. Not only are they not designed for this stress in general terms, but they have not had to work out the capacitance planning for it. A perfect example of this is transactional logging. When you enable transactional logging on source systems so that they can feed these monster cloud systems, that introduces an additional load across the board and these have not been planned for, not only are these not being catered to at an IT level , but they are often ultimately driven from other business areas and the normal slow business growth planning of humans.

The points to be learned from such problems are:

  1. Communication, communication, communication: With all of your existing people. You will find you will have communication issues, but this is not down to them not wanting to grow or change. This is simply down to the speed that the internal systems is set to grow is not aligning with the cloud systems expectation. You have to work with people from all areas. You have to get them involved from the minute you’re starting to do this. Try and give them a heads up on what you’re going to do. They often have to plan their budget and how much they spend on infrastructure change and people hiring a year in advance, suddenly turning on your AWS servers and cranking them up to 11 does not help them.
  2. Expectations: Most of your customers for your systems are now expecting consumer level speeds. All of your social media plans are nearly real time. Now, all of a sudden you are expecting that from human and IT Systems that are buried deep under a whole raft of procedures that you have not changed for decades. You cannot provide real time updates to such things with only changes to IT systems, you need to bring along the business systems that are supporting them. Explain this to your cloud system consumers so they can set their expectations along real human lines.
  3. Fear: A little bit dramatic, but its a real statement, people fear change, if you do not work with people on this they will fear these changes. To gain proper improvements from cloud systems you have to work with everybody, introduce them to the advantages and their place in such a changed world, help them grow as you are helping the IT systems grow.

The best analogy I can think of to explain all this is from the image that is headlining this post, from “Appleseed” a famous comic by Masamune Shirow which confronts a lot of the issues of change in the world. One of the doctors in the comic states “You can’t give someone a Cyborg Leg Just like THAT, It’d tear out when they try to run”. Think of your move to cloud in thoese terms.

Leave a Reply

Your email address will not be published. Required fields are marked *