Inside System Storage -- by Tony Pearson

Tony Pearson Tony Pearson is a Master Inventor and Senior IT Specialist for the IBM System Storage product line at the IBM Executive Briefing Center in Tucson Arizona, and featured contributor to IBM's developerWorks. In 2011, Tony celebrated his 25th year anniversary with IBM Storage on the same day as the IBM's Centennial. He is author of the Inside System Storage series of books. This blog is for the open exchange of ideas relating to storage and storage networking hardware, software and services. You can also follow him on Twitter @az990tony.
(Short URL for this blog: ibm.co/Pearson )
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (5)

1 localhost commented Permalink

Tony,<div>&nbsp;</div> I like the MPG comparison. A nice way of explaining what I was trying to get across in the various discussions regarding SPC relevance last week.

2 localhost commented Trackback

The value of a storage device is more than it's benchmarks- otherwise IBM would not offer it's considerably slower, but more versatile Netapps filer OEMed as "N series". The value of a solution (and through that, it's total cost of ownership) are more heavily based on business numbers like availability and manageability, and these are metrics that are hard to quantify.<div>&nbsp;</div> If you're comparing equal functionality in a system, then it can come down to benchmarks, but I think that's fairly rare.

3 localhost commented Permalink

Very true, but with something like SVC that sits in the middle of your SAN, we have to make sure it doesn't limit performance. As well as provide all the advanced copy services, data migration etc etc functions.

4 localhost commented Trackback

BarryB,Yes, we made sure.<div>&nbsp;</div> In addition to providing improvements in multi-pathing, data migration and advanced copy services, we improve performance in three ways: caching, striping, and load balancing.<div>&nbsp;</div> Caching - many business workloads respond favorably to cache, we call those cache-friendly workloads, and so more cache is often better than less cache, and providing additional cache, such as with SVC, provides additional performance. The key difference is that with SVC in front of a cached-controller, we get the multi-level cache effect, similar to having L1, L2 and L3 cache on processors. Perhaps this is a good topic for its own blog post, but until then, talk to one of your cache experts and have them explain it to you.<div>&nbsp;</div> Striping - this is the notion of spreading I/O requests out to different HDD. Both RAID-5 and RAID-10 benefit from striping performance effect. SVC can strip across RAID groups, across frames, even across different disk systems. Another good blog topic, but until then, talk to one of your RAID experts to explain it to you.<div>&nbsp;</div> Load-Balancing - some volumes, and some RAID arrays, deal with heavier workloads than others in the same box, or across arrays on the same data center floor. While EMC offers software to detect this and help after the fact, SVC chose instead to help eliminate or reduce this in the first place.

5 localhost commented Trackback

Open Systems Guy,Quite true. Performance is merely one dimension to measure and compare different storage offerings. We offer disk systems at all different performance levels, to meet the varying demands of the marketplace.<div>&nbsp;</div> In some cases, performance benchmark results may simply be used to identify the short list of vendors you plan to invite to an RFP. Obviously, IBM should be on the consideration list.<div>&nbsp;</div> In other cases, performance benchmark results may be used to focus on specific products. If certain products do not look like they will be powerful/fast enough to handle the workload, this may be a show-stopper, no matter how attractive other attributes may be.<div>&nbsp;</div> I agree some attributes are more difficult to quantify than performance. However, all attributes can be quantified to some level or another. Availability numbers are often quoted such as 99.999% which represents only 5 minutes of downtime per year. For manageability, it can be measured in how many steps it takes to complete a task, how many different programs or interfaces must be involved to complete a task, etc. I am glad to see that IBM's management software is ranked in Gartner's "Magic Quadrant" at the uppper right as one of the best available.

Add a Comment Add a Comment