IBM Chief Engineer Clod Barrera dropped in on theCube at IBM Edge 2012 (full video below). He shared a number of thoughts on the current state of storage, architecture, and technology in a brief but informative discussion with theCube panel. IBM has long been a leader in storage integration solutions and through the discussion it was clear that the company persists in that role.
Storage in the wild
Clod began with a discussion surrounding the nature of storage, in that today and as it has been at IBM for some time, storage is less insular, and that the company has been a sort of outlier in those regards. Going back ten years or so, most SAN discussions were self-contained in comparison to an end-to-end approach we see successfully being discussed right now. Storage is becoming more integrated into the systems stack. For example there are countless cloud scenarios where storage solutions are designed to solve specific problems. If you think about it, that is true, there are a number of categories you could put workloads into. Data Analytics has a high performance demand workload. Transaction or Database could be another. Collaboration and Web services, Archiving, Business Apps and Logic present still further categories. IBM meets these needs by focusing solutions around specific problems and that definitely includes storage architecture.
Another huge factor in the business today is the dialogue around analytics and whether infrastructure meets the goals of this ambitious undertaking. That introduces factors around facing customer issues to recognize the amounts of data at hand and the obligations to retaining it in a way that is affordable. Economics of owning data is one of the biggest risks to consider, and IBM’s solution sets are poised to answer those questions.
Barrera also touches on Big Data and whether it will finally deliver some of the solutions that business has required for so long. The past data models have not fulfilled the various promises that have come along with them, and Clod feels that Big Data as that solution is yet to be determined. The requirements are there- better speed, better scale, more value – all of these things will be increasingly complex elements of the big picture in alignment with the demands that our society places on technology. Not only will they have to be better, but they have to consistently be improving in significant ways. A clear case study for the value of analytics is in homeland security. The kinds of insights from data analytics can provide vast and significant gains in decision making in this particular field and improve many systems. Challenges in that space, as well as others would include issues such as privacy and the question of cost of keeping all this data.
Advances in data compression
The discussion then shifted to some of the real-time compression technology from IBM. In the past, compression was utilized only around cold compression, in reference to data that wasn’t truly active. The computational overhead in this construct was onerous, being limited by available technology. What IBM is doing now is doing real-time compression on online data as it is being received by the disk array. Apps, middleware, and so on are for all intents and purposes, unaware that this is even happening. Capacity therefore is increased by factors of 2, 2.5, even 4 times standard rates and can be ‘rehydrated’ just as quickly. This is a significant advantage that IBM wields to answer some of the questions that the presence of data represents.
Flash and the enterprise
Flash as a technology was not something that was picked by the enterprise. As Clod notes, it is not originally an enterprise level technology – It is prone to wearing out, it sustains too many writes, read and writes are unbalanced, and there are requirements involved to reclaim space. In order to get it ready for the enterprise, there is a good deal of big and important work to make the componentry worthy. In ten years, there is certainly an opportunity for another technology to supplant flash. It is likely that this will be driven by consumerization once again, and the goal then as it is now will be to making whatever that future of technology is as robust for the enterprise as possible.
As the conversation wrapped up, a few more technology concepts were covered. Again the answers to business needs are something that IBM tends to tailor to specific clients and specific needs. The object storage model was discussed, and the take away from Clod was that the time for this has arrived, that is right now with the needs for internet scale access, with individual level of multitenant security for people all over the world to get to data repositories. With a confluence of technologies to deliver the solutions for this need, the model can be simple, simple rest APIs, having security controls applied, and utilizing simple put/get interface. Compared to NAS protocols, the risk of exposure is greatly reduced. The use case is there and ready. These and other elements construct a road to the concept of unified storage, with object storage protocols in addition to what already is in place in block storage and NAS storage, the model would be unified in one place. In time, the eventual cost of flash will probably lend validation to an active data on high speed architecture and data tub on slow disk model.