«

»

Mar 06 2011

Adding an extra UCS Chassis

The architecture of the Cisco UCS blade system is fundamentally different from the others in the market place. There is no intelligence and no configuration build into the blade chassis itself. In fact, you won’t even find a management port on the chassis.

Instead, the UCS architecture is build around centralizing the intelligence in a pair of Fabric Interconnects (FI’s). These FI’s run the UCS Manager software which is responsible for all of the intelligence and configuration of (with version 1.4 up to 20) multiple UCS chassis under it’s control. This creates a ‘virtual chassis’ and adding extra capacity by either adding a new chassis and/or a new blade is extremely simple and fast as there is no configuration that needs to take place.

In talking to customers about their experiences with blades in general, the ‘blade 17 problem’ often comes up. Customers are happy with their blades and chassis until the moment it is full with 16 blades, and the next project arrives and the need for blade 17 (HP/Dell chassis hold 16 blades, IBM 14). Adding an extra chassis takes the same amount of work as the first chassis. The exact same configuration steps need to be run again and change windows need to be requested to connect to the LAN and SAN. You know what I’m talking about if you have these systems.

Cisco UCS doesn’t have the ‘blade 17 problem’. For one, the Cisco chassis has 8 slots and not 16, so one could argue we have a ‘blade 9 problem’. However due to the architecture where there is no intelligence in the Cisco UCS chassis – not even a management port – the situation is quite different.

But instead of talking about it – we decided to record our experience when we received an extra chassis for our Amsterdam demo lab. Our existing UCS system consisted of a pair of Fabric Interconnects (6120’s) and a single chassis. You will see all the steps required to unpack, rack, cable and power up an extra chassis. We recorded the time with an iPhone stopwatch to show you how we are doing in time. I did not cut out any footage, but did fast-forward where appropriate.

I hope you enjoyed the recording. As you might have noticed, we didn’t stop after powering up the chassis. We continued through the boot & discovery process and we assigned a service profile to a blade which we then booted to run ESXi. In this recording we did document the time for that but in editing I decided it was going to make the recording too long.

Right now I’m considering creating a Part 2 – where I focus on discovery of the chassis & blade. Are you interested in that Part 2?

1 comment

1 ping

  1. Mohammad Alammouri

    Nice video TJ … sure interested in part 2 !

  1. Adding a extra UCS Chassis – the configuration minutes » TJs Thoughts

    […] « Adding an extra UCS Chassis […]

Comments have been disabled.