Little Miss Enviro Geek : Hard Disks

Some personal news...

biwirth — Fri, 18 Sep 2009 03:24:00 GMT

Hello Dear Blog Readers,

(hmmm....that kinda sounds like the beginning of one of those SPAM emails)

I wanted to let you all know I am moving on from Microsoft and shall no longer be blogging here. I'll be moving an archive of these posts to my website http://www.greenitstrategy.com and shall continue blogging there on Green IT.

So what will I be doing now?

Well, I have decided to take some time off, write a book and maybe look into starting my own business. Regardless, you can always catch me through my website and please do feel free to visit me there.

Thank you to everyone who contacted me with their Green IT questions over the past two years and gave me feedback. I really appreciate your support and good luck with your own Green IT initiatives.

Kind Regards,

Bianca AKA Little Miss Enviro Geek

Great new whitepaper: Power In, Dollars Out: How to Stem the Flow in the Data Center

biwirth — Wed, 07 Jan 2009 05:11:47 GMT

First up - Happy New Year! :) I had a relaxing time (apart from the stupid oven cracking it halfway through the roast on Xmas day) and, when the mood struck me, I spent time updating my website with new features and content.

Back into the swing, in early December the Windows Server Performance <guys/girls/whatever> released a great new whitepaper with quite a lot of detail around optimisation of servers, as well as publishing their test results.

Some of the best practices covered include:

Shut down idle machines during off-peak times.
Opt for 2.5-inch instead of 3.5-inch disk drives.
Use low-revolutions-per-minute (RPM) disk drives where it is possible.
Select power-efficient processors and memory.
Install variable-speed fans and efficient power supplies in servers to reduce waste.
Use remote-controlled power strips to completely eliminate electricity flow to "powered-off" servers.
Ensure that Windows Server 2008 is configured to use the Balanced power policy.
Tune processor power management parameters to increase efficiency by up to 10 percent.
Consolidate workloads or combine server roles on idle and underutilized servers where it is possible.
Turn off hardware-based power management
Deploy the latest service packs and Windows Server releases.
Remove or shut down unnecessary roles, applications, and devices.

The 'shut down idle machines during off-peak times' section references a good article by Dave Ohara which provides some more practical suggestions for how you actually achieve this. However I do question the viability of turning off servers when they are not in use in **some** environments:

Issue number 1: I know that some of my customers have trouble even identifying what applications are running on certain servers and what the subsequent business impact would be if there were turned off. This sounds highly unlikely and slightly dark ages to some IT departments but I know from talking to others this is the case, especially on older hardware and OS' that 'just keep running in the corner'. This is not to say I condone not knowing what servers do - just pointing out that turning off idle machines is a more complex analysis process upfront for some.
Issue number 2: in regards to actually identifying some servers that don't need to be running at certain times of the day - I think this may be a nirvana situation in some less-than-leading-edge environments. Probably the 'easier-target' scenario I could see this working in is where you have a bank of servers, which Dave rightly points out - something like a Citrix farm in which each server may generally be configured identically and en masse (which would provide capability even at reduced performance until a new server(s) came up from being turned off, if that was necessary).

The whitepaper starts to get really meaty towards the middle of the document, covering some great detail on areas such as efficiency analysis, ACPI (C-States, P-States, T-States), interrupts, timers and test results for removing components to measure their power draw. A few good points:

Even when powered off, the test server consumed 27 watts - you might often hear a similar story along the lines of 'unplug your mobile phone because it draws power even when turned off at the point but is still plugged in'.
In the tests Memory consumed 54% of the total system power. Newer RAM consumes less power due to modern energy efficiencies.
Using 2.5" disks instead of 3.5" disks doesn't impact performance but does consume 50% less power.

Anyway before I give away the storyline, I'll let you go read the paper yourself: http://www.microsoft.com/whdc/system/pnppwr/powermgmt/Svr_Pwr_ITAdmin.mspx

Also check out a related blog post by the team who wrote this whitepaper, focusing on tuning power processor performance: http://blogs.technet.com/winserverperformance/.

Enjoy the theme music to go with the blog post: Endless Sacrifice - Dream Theater; Songdust & Delerium - I Mother Earth; King of Terrors - Symphony X :-)

Microsoft Research: Spin server disks down to save energy

biwirth — Wed, 16 Jul 2008 08:49:00 GMT

As part of ongoing research into the energy efficiency of data centers, Microsoft Research (MSR) have released a paper discussing the results of testing disk subsystems usage and potential to 'spin down' disks when they are not being used. In the past there was an assumption disks couldn't be spun down due to continuous use, however MSR found during their week long trial on 36 volumes in their data center there were "significant" periods of idle time.

Their solution to helping reduce energy used by disks, up to 45-60%, was to enable functionality called "write off-loading" - this allows write requests on spun-down disks to be temporarily redirected to alternative persistent storage (such as NVRAM or flash).

A few interesting facts from their paper:

enterprise storage sub-systems don't have advanced power management capabilities (because it is inherently difficult) and they consume approximately 10w when idle;
write off-loading is a block level, transparent process;
the key challenge was consistency - "Each write request to any volume can be off-loaded to one of several other locations depending on a number of criteria, including the power state and the current load on the destination. This per-operation load balancing improves performance, but it means that successive writes of the same logical block could be offloaded to different destinations. It is imperative that the consistency of the original volume is maintained even in the presence of failures. We achieve this by persisting sufficient metadata with each off-loaded write to reconstruct the latest version of each block after a failure."
They tested two types of off-loading: machine level off-load (writes are offloaded to a "logger" on the same server) and rack level off-load (writes are offloaded to any "logger" in the rack) and found the latter to utilise the least energy (55% of the baseline).

Check out the complete paper outlining the details of the tests and associated research here: http://research.microsoft.com/camsys/paper-final.pdf

Little Miss Enviro Geek : Hard Disks

Some personal news...

**Great new whitepaper: Power In, Dollars Out: How to Stem the Flow in the Data Center**

Microsoft Research: Spin server disks down to save energy

Great new whitepaper: Power In, Dollars Out: How to Stem the Flow in the Data Center