Bladewatch.com

An interesting article talking about how IBM and the Syracuse University are working on building one of the worlds’ most energy efficient data centers – very cool, it’s great to highlight the concepts and discuss, the best practice and the range of technologies or processes used to achieve this. Do check it out, they’re using a range of solutions to achieve this.

Looking at how we power the data center, looking at the different elements of the infrastructure, the server, the storage, the network as well as the data center in terms of air flow (checking that we have effective and efficient air flow to where it’s needed) as well as effective power distribution using efficient power supplies and possibly DC power? It will all depend on where you are in the data center life cycle.

It’s a multi feed activity, by that I mean, we should be looking at the infrastructure, the application and the usage patterns:

• Data center design – power/air flow and environment -  what operating temperature should we run at? Could we run it hotter and use less power for cooling
• Server/storage/network – can we use more efficient hardware, using more efficient power supplies, lower voltage processors, solid state or the smaller more efficient drives/SAN boot
• Infrastructure configuration – can we power down what we don’t need, can we remove those organizational screensavers to reduce idle cpu utilization, can we configure the infrastructure more effectively – scale down the network when it’s not needed, remove redundant power supplies from servers that don’t need them. This could include drivers/firmware and best practice configurations or settings being applied.
• Application configuration – can we configure the application to exploit more of the processor functionality to gain performance gains, or could we use graphical processors? Could we better code or optimize the code for the platform?
• Infrastructure utilization – can we use technologies like Citrix, virtualization and grid to create shared infrastructure pools which are more heavily utilized, a citrix farm hosting applications rather than separate servers per application, server virtualization to achieve more cpu and memory utilization. Grid to create a shared compute model where we only consume what we need to use.
• Storage efficiency – do we need to keep all the data we have online right now on our disk shelves – can we use data deduplication technologies, offline or nearline storage where we archive off data and improve data utilization? Can we not do more to reduce physical storage consumption?

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.