I’ve just spent the week with a customer exploring the operational differences that UCS has over any other data center compute infrastructure, and the result: it’s incredibly powerful, so you need to be very careful about how you use those powers.
This post explains three major features of UCS that just aren’t possible with any other system. Period. But they are very powerful, so use them carefully.
The first Spidey Power is Firmware Management capability. Gone are the days of “update the firmware when the new server is installed, and then never again”. Let’s be honest, firmware management of HP, Dell and IBM is notoriously difficult and hard enough to stop people doing it unless they really, really need to.
With UCS, you create a package for firmware – to update the BMC (ILO) and the server BIOS, adapters and controllers – and you simply attach that to a Service Profile. Then, when that Service Profile is associated to a blade then UCS will, automatically, upgrade the firmware on the blade. Even better, attach the firmware policy to a Service Profile template and every new workload gets the latest firmware. That’s incredibly powerful on its own, but it doesn’t stop there.
If you get a new firmware release, create a new package - but how do you apply this to existing Service Profiles associated to blades with running workloads (e.g. vSphere with VMs)? You can just attach the new package to existing Service Profiles when you are ready to make the change, and UCS will power down the blade, update the firmware and bring up the workload again.
But here’s the responsibility bit: if you don’t do this methodically and under change control, you will bring down running workloads and cause service disruption. Great power = great responsibility.
The second Spidey Power is Updating Templates. You can create a Service Profile template that is used to create Service Profiles with consistent configurations, but with an Updating Template you can propogate any changes to the template to all running Service Profiles. Very, very powerful. Mix this with the Firmware Management package, and you can immediately update all Service Profiles via just one change to the Updating Template. Wire once and walk away! If you have a system wide change window, this is awesome. If you do this by mistake, not so awesome. Process is key here. Our customers are smart enough and big enough for these powerful systems.
The third Spidey Power is the Policies. There are 25 policies in UCS that you can attach to Templates or Service Profiles to ensure consistent configurations. If you change policies, then they cause changes to the Service Profiles that attach them. Take Adapter Policies that tune the blade adapters, such as Tx Receive sizes. If you change this setting then all Service Profiles will be rebooted so that UCS can customize the adapter for the new settings. Fantastic power, if you know what you’re doing.
As I pass the keys to the new UCS over to the customer, I know that they’ll soon be exploiting these new features and Pandora’s Box will be well and truly opened. Their old IT infrastructure’s days are numbered.