Next-Generation Storage: Low Cost Isn’t Enough
New vendors trumpet products with lower price tags, but if they lock you into a single source, what’s the point?
Recent travel has found me in Europe and the U.S. speaking at events sponsored by reseller/integrators that specialize in storage technology. Often led by former industry insiders who are seeking to “redeem their souls” after working for so many years in the marketing and sales organizations of “monolithic Big-Iron” storage vendors, these firms are now attempting to “right old wrongs” and combine knowledge of legacy and leading-edge technologies to help clients fit the right storage solution to the needs of their businesses.
One of integrators is infoStructructure Enterprise Storage Solutions of Denver, CO. I enjoyed participating in a small event hosted recently by CEO Greg Wilson that attracted about 40 attendees from companies that account for 80 percent of IT expenditures in the region. The event also featured vendors with which Wilson has chosen to partner, including Onaro, Isilon, 3PAR Data, and EqualLogic — their presentations, about 45 minutes long, mostly described their products.
Overall, it had the look and feel of a half-day conference you would expect from a professional events team for a big league vendor. Greg, however, and his two-person team, put the show together as a labor of love. It was their coming-out party and Wilson had the predictable anxiety of a newbie: Would anyone show up? Would the event shine a good light on his venture? Would the sponsors regard the show as money well spent?
He shouldn’t have worried. Setting the stage with a five-minute overview of his recently-founded firm, his presentation included a declaration of his firm’s core ethics. He promised to do the best he could to solve customer problems, even if the solution did not include technologies re-sold by his firm. So sincere was his pitch, I think he had the entire audience in the palm of his hand.
There is nothing cynical about Wilson. He genuinely believes that he will be able to do right by his customers using his knowledge of available products. I hope that he can, but I will watch this story as it develops. What makes fulfilling his vision problematic is the list of products that he resells.
Setting aside Onaro, which provides an interesting storage-management and capacity-planning software solution, and Isilon, which focuses on clustered NAS (more on this next week), the rest of Wilson’s play card consists of products aimed at displacing monolithic storage with less-expensive monolithic storage.
I'm sure I will get an argument about this characterization from both EqualLogic and 3PAR, but I am going to stick my neck out anyway. I listened intently to their technical presentations and walked away with the feeling that each vendor, in its own way, is merely seeking to be EMC’s or HDS’ or IBM’s “mini-me.”
I listened intently to 3PAR representatives as they spoke about their “post switch architecture.” (If you think that moniker smacks a bit more of marketecture than architecture, you should know that I found myself dizzy from all of the similar quasi-tech expressions used by the presenter during his talk.) 3PAR is a group of 10-drive magazines united under a meshed set of cluster heads. There are two flavors: the S400 (with a backplane supporting 1.4 GB throughput) and the S800 (supporting 2.8 GB throughput). The vendor says the more nodes you deploy, the faster the machine becomes—more drives, more I/O.
3PAR uses a unique method for parsing disk space for use by applications. The product creates “chunklets” of drives that can be aggregated to create virtual volumes on the fly. It also uses custom ASICS for data movement.
Software on the machine includes RAID 10 (“mirror and stripe”) and RAID 50 (“parity and stripe”) plus utilities for virtual copy and just-in-time space allocation (“demand on write”). From what I gleaned, it ships with a point storage management product.
So we have a storage target that offers good performance (Oracle testifies to this, which is not surprising because Oracle is a “multi-time investor” in the company, we were repeatedly told) and has a solid scalability story to tell. I discounted the former, especially the Oracle endorsement, since 1) Oracle already supports writes to disks of different block architectures and doesn’t need special accommodation from 3PAR to support this; 2) Larry Ellison has also made a significant investment in Pillar Data Systems, which would seem to be a direct competitor to 3PAR; and 3) Oracle would be expected to say positive things about the performance of any platform it had certified to work with its wares.
The Lock-In Factor
The second differentiator, scalability, didn’t excite me very much either. Here we have a scaling story that only works if you buy only 3PAR drive magazines. How is this different from the scaling story of EMC, et al? This sounds suspiciously like a vendor lock-in strategy—even if it is purchased at a price point “substantially lower than a comparable EMC solution.” Lower acquisition cost doesn’t make a lot of difference in the ultimate cost to the consumer of lock-in technology in terms of dependency on a single vendor and lock-out of innovative alternative approaches.
EqualLogic did a somewhat more interesting presentation, citing at least some hint of a business case for its products. Quoting analyst Bob Passmore, the speaker said we could expect an accelerated adoption of iSCSI-based storage to cope with the data explosion, based mostly on its lower cost. They have provided scalable “frameless architecture” for block storage, based entirely on standard iSCSI, since 2003, selling more TBs than any other company but Network Appliance. NetApp only tops the most iSCSI sold charts at IDC because it ships iSCSI readiness in all of its products. IDC counts this as a sale whether folks use the NetApp iSCSI target functionality or not.
So data is growing, iSCSI is a popular standard, and EqualLogic sells a lot of it. The product also offers “set-and-forget management”—a comparatively simple configuration as a RAID 5, RAID 10, or RAID 50 clustered array with configuration wizards and a simple Java-based element manager. (The poor fellow tried to demonstrate simple configuration, but his demo wouldn’t launch.)
Again, there was no problem with the presentation as far as it went, but it left me wanting more. He noted the availability of Serial SCSI (the X-Series) and Serial ATA (the E Series) versions of his product, but never explained what applications might require one or another. He also spoke about automated provisioning: disk pools would create themselves automatically, in response to load characteristics. The later sounds interesting and is something I would like to test in my labs sometime.
All in all, however, the solution is a bit like 3PAR's. We lack a clear understanding of the specific application problems that it solves and we are locked into the vendors' own disk controllers and spindles.
Neither vendor talked about the mark-up they were placing on their commodity hardware-plus-custom software suites. I was certain that the cost of the former was being driven upwards by the latter and wanted to get an exact read on when these products would turn into hardware agnostic software offerings that could be used with any disk. Such analyses were not forthcoming.
In the end, I saw CEO Wilson, armed only with these products, offering basically the same value proposition he had once proffered in the service of his previous employers: one-size-fits-all architecture, potentially expensive product lock-ins, with a one-throat-to-choke contract. I think that if he is to live up to his idealistic code of ethics, he may well find himself searching for some additional wares. To serve his customer well, he needs to “integrate” and not just “resell.” He needs to “purpose-build” storage solutions for his clients that manifest best-of-breed capabilities at affordable prices.
Greg is a good guy and part of a new generation of storage resellers and service providers that seem to be trying to distinguish themselves from the purveyors of monolithic Big Iron. I have little doubt that he will build up his play card in time to avoid falling into the habits of his previous life in the industry. I wish him every success.
Your comments are welcome. firstname.lastname@example.org.