By Calvin Zito, @HPStorageGuy
If you rather download this as a podcast, right click here and save the MP3 file. You can also subscribe to my podcast on iTunes - go to the iTunes store and search for "Around the Storage Block" or you can open this link and click on the "View in iTunes" button under my picture.
Greg Schulz is an analyst/author and I got to do a podcast with him a couple of months ago at SNW where we talked about a session he had focused on cloud computing. He has a new blog he's writing for InfoStor that he's calling Storage Metrics and Measurements that matter. In his latest post, he takes up the HP 3PAR SPC-1 benchmark results and addresses the question of "did HP short-stroke the SPC-1 results".
I'm happy that Greg wrote an article to clear this up. There's a bit of background I need to give you here so bear with me. A few months ago, Jon Toigo wrote a story for Storage Magazine where he wrote, "...I have to chuckle when HP/3PAR carries on about having the fast hand in storage at 400,000+ IOPS. Short-stroking a lot of spindles does buy speed, but at a non-trivial expense in terms of power." (See IOPS per what? for the rest of the article).
Short stroking a benchmark is basically when you throw a lot of spindles at a workload and use a small amount of the total capacity to achieve the results; data on the drives is confined to small physical space so the disk drive heads don't have to travel far to grab the data off the media, reducing disk latency and thus the term short stroking.
How Toigo came to the conclusion that we were short-stroking is - well - a question for him to answer because I don't know. If you open the SPC-1 executive summary you'll see that the unused physical capacity is less than 14.5% - clearly not a case of short-stroking. Jon to his credit did include a post script correction on an unrelated article he did for SearchStorage and also wrote a blog on his site correcting his error. Unfortunately, not many people saw his blog correction as the question comes up more than in should.
If you want to see short stroking in action, take a look at the recent SPC-1 submission by NetApp where they had an unused capacity of 43.2%. The maximum allowed by SPC-1 is 45%. As I already said, contrast this with the HP 3PAR submission which had 14.54% unused capacity. I don't know enough about what NetApp did to be sure they short-stroked their benchmark but they sure had a lot of unused disk capacity.
Anyway, all that said, Greg's blog on InfoStor is worth a look and he has a short follow-up on his blog titled "Give HP storage some love and short strokin". Virtual man hugs to you Greg!