We’ve been championing the tightly coupled storage and computing model for a long time. When it was unfashionable, when it was discarded as “this is something you should not do” by others “who knew better”.
Now, the ideas, the concepts, the thoughts, the designs and implementations behind it are all around. Joyent’s Manta system is an implementation of the concept. Arguably, the more advanced MapReduce and Hadoop designs are also implementations … have the data right next to the processing, and provide gargantuan bandwidth locally to the data.
Make the fabric insanely fast. Remove bottlenecks. Remove impediments to data flow. Architect so that you can scale out using some of the fastest storage and computing units available in market.
But to do this right, you need very fast individual boxes. You need machines and appliances that can push absolutely insane data volumes, and perform processing in-situ, and then you can distribute the processing and the data across many such systems.
It was only a matter of time before we came on the threat radar. Showing up once with a very fast box, setting many records, could have been a fluke. So we did it again. And set more records.
It would be fair to say that our Cadence Time Series Analytic appliance may be one of the, if not the, single fastest tightly coupled storage and computing box in market today. We’ve got the benchmarks to prove it.
And since we did it twice, we are now a threat to the established order.
In NYC last week, at the STAC summit, I watched as one of our erstwhile competitors did a poor job of trashing us (without mentioning our name), actually making our argument points in the process.
Today in Chicago, also at the STAC summit, I watched this same erstwhile competitor (different speaker though), again try to trash “single box” direct attached storage, hand wave away record setting performance. Again, in the process, someone asked a set of great questions which beautifully made our points for us.
Someone asked whether or not it made sense to build parallel file systems out of the fastest possible building blocks. The panel agreed, apart from the erstwhile competitor. Someone asked if using many mount points for petabyte scale systems turned the problem from performance management into a data management problem. Apart from the erstwhile competitor, I got the sense that the other panelists (and audience members around me) agreed.
It is always dangerous to underestimate a competitor. I have no illusions about these folks, their revenues are easily 100x ours (guessing at this), they have many more people, much more brand recognition. We are puny in comparison.
But we have the fastest box in this market. And we keep coming back with faster systems. How can this be? Haven’t we hit our limits? Aren’t we going to go away soon?
[insert big evil grin here]