In this post, I’d like to focus on what has changed over the last 10 and 20 years in Capacity Planning and its challenges. Make sure to read Capacity Planning Part 1 before continuing on. I will start by covering couple of very important issues that Capacity Planners have to deal with.
Hardware used to be very expensive and it used to be a large part of any IT budget. Today that is very different. Hardware cost has come down. Instead, Staff, Software Cost and Utilities have increased as a percent of IT budgets. We used to pay more for license as we would upgrade our hardware. But that was a smaller increase. And that had a lot to do with the head room customers would plan and pay, acquiring hardware. More and more Software vendors are charging by Socket, Core, usage based pricing, & every other way that is possible to increase their revenue and trying to get their fair share. As customers add and upgrade their hardware, they also end up paying more for Hypervisors, as well as all software running on all that additional over configured hardware that is usually being wasted. More and more customers are paying attention to additional Software License Cost every day.
Moore’s Law has reduced Hardware Cost. And it has increased number of Hardware Vendors and all the options they offer. I see a lot of customers try to make their life easier by picking and sticking with one, or at least less number of hardware vendors. The relationship with this One, or few Hardware vendors and customers get better and better. And most offer more help in Capacity Planning. Hardware Vendors are playing a larger role in selection process of hardware acquisitions, and also in hardware refresh. Three year leases are more and more popular. And some customers actually let the hardware vendors to do the capacity planning for them, and come and replace a large subset of their hardware on some regular basis. And there is also value in sticking with same vendors and support, especially if customers are happy with it. This is influencing the larger head rooms that customers have today vs a few years ago.
More options we have, more complex things get. I recall CPU being the main bottleneck and most expensive part of hardware. And strange enough, these days more and more customers are challenged with buying the right amount of Memory. Some add Memory & CPU on regular basis. And some buy fully loaded servers with all the CPU and Memory they come with. But one more challenge has been added for the Capacity Planners. I can see that no matter how good of efforts you put in, you will always buy one resource more than you need and one resource less than you need. But measuring and predicting Memory usage is harder these days. As we all know, our Databases and Application Servers allow us to allocate and reserve memory. Operating systems will not steal the memory from the process. The more we allocate, the more they will be used, and some very inefficiently. This makes it hard to know how much memory you really need. And collecting just memory allocation and usage is not as accurate as we like.
Please provide your feedback as I like to hear your perspectives. I always learn from my customers and peers.
More to follow soon...