By Michael Callahan
As my colleague Calvin Zito mentioned earlier today, I wanted to make a few comments on EMC's announcement of the EMC Atmos (nee Maui).
Since we first started talking publically about the ExDS9100 system in May, I've been struck by the incredible diversity of the software layers that people want to build on top of it. In fact, quite a bit of this has been a surprise. Much of the ExDS design grew out of our many years experience delivering software into environments very much like those which Atmos seems to target: Web 2.0 companies, telcos and service providers. Based on that experience, we put a lot of emphasis on building something that could easily fit into existing environments (with standard racks!), be extremely simple to manage and, of course, be very cost-effective-while still offering a lot of flexibility in both hardware and software. On the hardware front, a single ExDS9100 node scales from four blades to sixteen, and from roughly a quarter petabyte to over three-quarters (in just two floor tiles) -- and a customer can choose whatever combination of performance and capacity is desired, and even change the configuration of a running system by adding blades or capacity blocks.
But, as I said, what's particularly surprising to me is what has been happening on the software front -- how valuable it has been to offer flexibility there. We know that EMC with Atmos is taking a proprietary, monolithic approach to the software layer that provides distributed replication, location services, and so on, on top of the basic building-block of local high-density storage -- but we think that's a mistake, and not what the market is looking for. The last thing our customers want is to be locked in by a proprietary software layer that limits their technical options and ties them to one vendor.
We have a whole set of customers-admittedly, the ones in the vanguard of these next-generation cloud applications -- who already have commitments to existing software that perform functions like these, whether built in-house, open source or commercial third party. But even apart from these leading customers, we're finding that in other verticals, such as bioinformatics, oil and gas, and media, customers see tremendous value in the density, manageability and scalability of the ExDS system but are looking for very specific solutions with software stacks that are customized for their particular application needs.
For example, Calvin mentioned our partner Ocarina Networks. With Ocarina, we've built a media archiving solution for a major film studio that allows it to maintain an online archive of its digital assets, using Ocarina's advanced content compression and optimization algorithms that include specific tuning for the data formats used in the industry. Similarly, we've seen bioinformatics and oil and gas customers looking to bring parts of the software stack down into the storage layer, in ways that provide specific value to their applications.
Now, there are customers looking to adopt something in this space for the first time, and we'll work with them, including offering an HP-provided option -- but the key point is that our approach is based on supporting the customer's desire at once to have a market-leading storage system that provides unparalleled terabyte density, power efficiency and operational and acquisition cost effectiveness and at the same time let them retain the flexibility to build an appropriate software stack on top of that, that doesn't lock them in to one vendor.
We'll be talking more about how other customers are using the ExDS9100 in the coming months, but for now I'll just say that I think this emerging market for capacity-optimized storage is one where there will be more diversity and innovation, rather than less, and where customers will have more ability to customize their environments for their own needs -- and will not have to place all their data into some vendor's opaque cloud.
Michael Callahan, Chief Technologist
HP StorageWorks Scalable NAS Division