In many levels of government, the mandate has gone out: the cloud is where IT should look first when hosting new applications (or re-hosting existing systems). There are good reasons for the mandate, but that doesn’t mean that there’s no longer a place for the traditional on-premises server in government IT.

What it does mean is that the servers you add to your architecture need to be carefully configured to optimize their performance for the critical purposes they serve.

What kind of features should you be looking for in the servers you still need? Three things will be important no matter what specific purpose the machine will serve. Be sure to look for:

  1. Cores (the number)
  2. DRAM (the size)
  3. Storage (local only)

Cores: Not so long ago, if you wanted to get a high-performance server, you would look for a multi-processor system. You might still look for more than one processor, but even more important is the number of processing cores in each chip.

In many circumstances, a single 8-core processor will fuel higher performance than twin dual-core chips. Once you get to multiple 8-core chips, you begin to look at serious performance upgrades. And a significant benefit to the multi-core architecture is that you’ll be getting the performance boost with far lower electricity consumption than would be the case with all those processing cores housed in individual packages.

Count the cores, and you’ll understand basic performance characteristics.

DRAM: If you know that a particular server will be used for one specific application and nothing more, then you can size the memory to the software vendor’s recommendation and be safe.

If you’re going to be using the server for multiple virtual systems and the applications that may come until the next hardware refresh cycle, then the general tip for RAM is “go big.” It’s rare to hear a manager say, “I wish we had bought less memory for that server,” while many long for more memory (or the add-on budget to top off the memory in an existing server).

Buy bigger, and you’ll likely have faster applications and happier users well into the future.

Storage: Server memory can be a complicated topic all its own, but here’s a tip for just about every server configuration you’ll put together: specify a fast SSD for the boot volume with enough capacity to hold the basic system images and essential application code.

Data storage can live in its own world, and you can get as wild as you’d like with that; but if you keep the system code on fast storage, you’ll boost the performance of every application instance and virtual machine.

If you pay attention to these three components, you’ll create a server that can be a successful part of any hybrid architecture. You might even find that the right on-premises server will improve the performance of those cloud components that policy requires. And that’s a win for everyone.