A few months ago, EMC made arguably its biggest announcement in the last 10 years--the Symmetrix DMX family. Performance was touted as industry leading. The company claimed the architecture to be revolutionary--with theoretical throughput levels soaring beyond anyone on the planet. After being beaten down on the performance issue for so long, the company no longer had to dance around it.
I didn't make performance the issue, they did. I said, "I have no reason to doubt EMC's claim." I also said, "If this pans out, this box is years ahead of the competition." I meant it. So, imagine my surprise when we decided to put the DMX and Hitachi Data Systems' hottest box head to head in a multilevel performance bake-off and EMC said, "No." Yes, they said no.
Enterprise Storage Group (ESG) has the hard-earned reputation of being the industry pundits that stand up for the rights of the common IT man. We received more than 500 individual inquiries about performance alone after the EMC announcement--to which we replied along the lines of: "It sure looks impressive, but we haven't been able to test it." So, we had a brilliant idea--let's test it.
We've started an independent test group we call ESGlabs. It's comprised of analysts, performance gurus and a committee of "real" IT professionals from both the storage and the application/database world. We'll use the independent test lab at Imation, and test help from GlassHouse Technologies. We understand that benchmarks are just that--not real world in every case. We also understand that most benchmarks are easy to manipulate by vendors. We won't allow that, of course.
Our objective was simple. Create an RFP that would emulate a real-world set of environments: a mix of realistic workloads such as CRM and other database-centric applications and end-user file traffic, connected through popular switches and representing multiterabyte environments. Sounds like your shop? That's the whole point.
We want to pound on the box and see where it dies and under what conditions. We also would find out how much that configuration costs--really costs. Quite frankly, I was worried about HDS participating. I figured they really had the most to lose. Without hesitation, they said, "Anywhere, anytime." For that, I give them credit.
At the end of day, if you make performance an issue, you need to give the user some validation points. We can't use real installations (on the record), because EMC makes people sign non-disclosures that are very intimidating. While 90% of the time performance is already "good enough," 10% of the time it isn't and performance matters.
So, while I continue to respect and admire the company 99% of the time, this is the 1% of the time where I just want to shake some sense into the giant. This is how the EMC of old acted--and just in case you forgot--no one liked the EMC of old. Why should the IT guy trust EMC's other claims about how it's changed (that are all probably true, mind you), if the company isn't willing to prove it on this?