Comments for Data Storage, Stategic Planning & Systems Development products and services | Silverton Consulting

Comment on SCI 2011 May 27 Review of latest SPC performance results by OCZ's new Octane SATA SSD pushes latency limits below 100μsec | RayOnStorage Blog

Thu, 20 Oct 2011 20:57:03 +0000

[...] latest RAMSAN SPC-1 benchmark results, under ~90μsec measured at the host level! (See my May dispatch on latest SPC performance). On the other hand, how they measure 90μsec host level latencies without a logic analyzer [...]

Comment on SCI 2010 Jan 28 last ESRP v2.1 (Exchange 2007) results analysis – 1K-to-5K mailboxes by Latest ESRPv3 (Exchange 2010) results analysis for 1K-to-5Kmailboxes - chart of the month | RayOnStorage Blog

Mon, 19 Sep 2011 23:04:08 +0000

[...] following a trend first shown in ESRP v2 (Exchange 2007) results. We reported on this in our January ESRPv2 analysis dispatch for this year [...]

Comment on SCI’s Newsletter by HDS buys BlueArc | RayOnStorage Blog

HDS buys BlueArc | RayOnStorage Blog — Thu, 08 Sep 2011 18:19:50 +0000

[...] In the mean time, scale out NAS products continue to move forward on a number of fronts. As readers of my newsletter know, currently the SPECsfs2008 overall performance winner is a scale out NAS solution using 144 nodes from EMC Isilon (newsletter signup is above right or can also be found here). [...]

Comment on EMC 2009 Aug 25 CLARiiON and RecoverPoint announcements by How aggressive can you get on datacenter power usage? - NEW RESUME PORTAL – NEW RESUME PORTAL

Fri, 26 Aug 2011 12:36:49 +0000

[...] (IPM). One of a storage professionals we demeanour to for good information, Ray Lucchesi has some records on hoop spin down on a renouned storage product. This of march is for a tier of storage that can catch this form of [...]

Comment on SCI 2010 Nov 18 Latest SPC results by Administrator

Administrator — Thu, 24 Feb 2011 20:33:35 +0000

KD,

Thanks for your comment and good catches.
- Yes, I missed the use of the Flash Cache on the FAS3270A, and have removed it from my next SPC update on this chart.
- The #3 Xiotech ISE was labeled wrong, it was actually the Xiotech with 20-300GB drives.
- The New IBM DS3524s also look pretty good in my latest SPC performance update (due out late this month to my newsletter subscribers).

Sorry, I didn’t see this earlier as I hadn’t set the dispatch posts to use intense debate. I will fix that shortly and should be more responsive in the future.
Ray

Comment on SCI 2010 Nov 18 Latest SPC results by KD Mann

KD Mann — Wed, 09 Feb 2011 16:41:12 +0000

Hi Ray,

This is great stuff — thanks especially for highlighting the IOPS/spindle aspect. As you know, depending on the choice of array vendor, the same workload might require twice as many spindles using an inefficient array.

Couple of notes however;

You said the IOPS/drive chart “excludes systems that use SSDs or Flash Cache”. I note that the NetApp FAS3270A uses 2x Flash Cache ‘PAM’ modules (~$100K worth) and should therefore be excluded here. NetApps best previous result was only 224 IOPS per (15K) spindle, making NetApp one of the least efficient architectures for this kind of workload. This appears to be the result of their RAID-6 implementation and the “write penalty”.

2) The Xiotech ISEs with 600GB disks (showing at #2 on your chart) tested at 12,603 IOPS from 40 spindles (315 IOPS/spindle), so it should be much further down the list.

Finally, a couple that are not on your chart yet are;

(1) the Xiotech model tested with 20x300GBx15K drives – impressive performance per spindle, but at $28,000 per TByte (list price for 2.3TB usable capacity), it costs as much as Flash SSD!?!?!

(2) the new IBM D3524 (LSI 2600) results look fantastic. 308 IOPS per spindle (using 10K disks and 94% of disk stroke), at $7,000/TByte and $3.26/IOPS — both are list price (no discounting tricks). I don’t think I’ve ever seen an SPC-1 result that clobbered both cost/IOPS and cost/Capacity metrics on the same test.

Looking forward to your next update…thanks again.