In-Depth

Storage Innovators, Part 2

Sometimes innovation is as simple as reviewing, and replacing, a vendor service and support contract with one from a qualified third-party provider.

In my last column, I noted that innovation in the storage industry might not come in the form of a product but rather as an ecosystem that enables us to divorce pricey and underutilized "value-add" software from storage arrays, enabling lower costs for the hardware kit while permitting us to buy only the functionality that we need from best of breed vendors (see http://www.esj.com/news/article.aspx?EditorialsID=3364). I argued that such a change from the way that storage products are shipped today would help companies purpose-build infrastructure -- right-sizing them from a budgetary standpoint and customizing them to actual business and application needs.

The prerequisites for such a strategy were linked to integration and manageability over time. I suggested that an under-discussed aspect of W3C standard Web services, consisting of direct or indirect calls by applications to infrastructure for the services and resources they require, would potentially solve both the issues of integrating network hosted value-add functions and managing the underlying boxes and plumbing in storage.

My inbox filled with comments from vendors who said they were pursuing such Web services enablement in their products. Some, like Xiotech ISE, are shipping today. Other Web services-enabled products will ship in the coming year.

Innovation in storage doesn't have to come in the form of a new box or new software for the box. In fact, with the current economy, real innovation comes from strategies that fulfill the three components of business value: cost-savings, risk reduction, and top-line growth. Two innovations I want to cover are storage virtualization and hardware service and support.

Storage virtualization, on its face, has a no-brainer business value case to offer. Theoretically, the perfect storage virtualization approach takes gear from different vendors and installs it under a common blanket so it is easier to manage from a capacity standpoint. The unfortunate aspect is that value-add features on arrays have made it difficult to virtualize un-like arrays in common. Hitachi Data Systems' Universal Storage Platform (USP) endeavored to create virtual storage infrastructure several years ago by placing a switch in the head of the array and enabling consumers to hang arrays from HDS or other vendors off the back of their box like so many JBODs. The USP head would then provide common management across all arrays while presenting servers with virtual connection ports to access the back end disks.

Essentially, this approach gutted the value-add software on the third-party arrays that you connected to the USP head (so said EMC at the time of the USP announcement, arguing that what you paid good money for when you bought standalone EMC arrays would be lost if you embraced such an unsupported configuration) and placed it in the USP itself. My take was that this approach could work for many shops, though in the end it was really just a reallocation of stovepipe functionality to another hardware vendor.

At the opposite end of the spectrum was virtualization provided at the server side using products such as then-Veritas' storage virtualization software, Volume Manager. Volume Manager worked to enable LUN aggregation, but it was a bear to maintain across hundreds of servers.

The third option was to provide virtualization on a switch or appliance, but this was usually limited in its efficacy only to storage interconnected via Fibre Channel plumbing. Moreover, it could be used only with gear from a single vendor: carving and splicing volumes from different vendors' gear would create instability in the resulting virtual volume.

One company that seems to have broken the mold is DataCore Software in Ft. Lauderdale, FL. I have covered them in this column many times over the years but I see their wares today as offering even greater value in the context of the recessionary economy.

DataCore virtualization products support multiple plumbing options and CTO Ziya Aral says that he has managed to circumvent most of the peculiarities added by box vendors to their gear that limit the ability to slice and dice their storage in a heterogeneous pie. A combination of SANSymphony and SANMelody have surmounted most of the proprietary hardware and plumbing issues, says Aral. "When we use the storage, it has already been created in stable LUNs."

The gating factor in DataCore virtualization has been speed: the slowest storage in a virtual volume set the speed for the overall construct. This will be addressed shortly by a new technology code-named "Virtuoso," says Aral.

Additionally, Aral is keen to work with Xiotech's ISE Ecosystem to avail his software of W3C standard hooks. "I see a standardized method for requesting services and resources from infrastructure as an approach that everyone has been hoping for for years. The ISE Web services team at Xiotech is the best in the business and we want to build on their expertise and ours to making automated provisioning and management a reality."

Aral and George Teixeira, CEO, added that customers confronting the economic downturn have already begun realizing the cost advantages of virtualizing storage with DataCore products. A major airline, Aral recalled, deployed DataCore wares to virtualize IBM Shark arrays, then used the technology to move toward less expensive hardware platforms. DataCore Software enabled them to right-size storage costs by moving value-add software off the array and into the network.

Said Teixeira, "We have been seeing this sort of thing at every shop where our virtualization solution was deployed. Last week, I was with a client in Colorado Springs that had been paying $10K per month in software licenses and service and support contracts to EMC for a 2 TB storage solution. I couldn't believe it. The integrator there switched them to DataCore for storage virtualization, high availability, and management, got rid of the EMC gear, put in a much-less-pricey and higher-capacity solution and virtualized some of their servers, replacing the server gear there and deploying Virtual Iron, for a total price of $60K. Not only did it dramatically reduce costs for licenses and maintenance, the company was also able to reallocate two people to other work."

Service and support is a key driver of operating cost in contemporary storage, Teixeira correctly observed. For an increasing number of hardware stovepipe vendors, the revenues realized from gear sales are almost inconsequential compared to what they expect to realize from service and support contracts over the five to seven years that the consumer owns the box. Some resellers have told me that the difference is just another manifestation of the razor/razorblades dynamic, and one that explains the generosity that some consumers are finding in the hardware discounts that are suddenly being offered by their preferred name brand vendor.

In truth, the service and support contract represents a "second-order cost" that may not be scrutinized by a consumer, according to Mike Linett, CEO of Newark, DE-based Zerowait, for its full impact on OPEX costs at the time of equipment acquisition. Linett is correct.

Recently, when asked why a small 1 TB upgrade to an existing array carried a hefty $53K price tag, an HP representative explained to one of my consulting clients that the modification also required the expansion of the customer's service and support contract coverage. "For $53K," the sales rep said, "you don't just get a couple of disk drives, you also get all of the love that goes with it." I can't reprint what the customer said in response, but suffice it to say, such a sum struck him as an awful lot to pay for love.

Linett says that such stories come with an increasingly familiar ring. His high-availability storage engineering company provides third-party service and support for storage gear, including NetApp Filers, at a fraction of the price that the vendor charges. One of his customers, a newspaper publisher, Linett says recently told him that using Zerowait service instead of NetApp service saved the company over $100K per year.

Sometimes innovation is as simple as reviewing, and replacing, a vendor service and support contract with one from a qualified third-party provider. Says Linett, "The customer wins either way. Telling NetApp that you are considering replacing their service and support agreement with one from Zerowait will likely get you a huge discount on the contract that NetApp is offering."

Your comments are welcome: [email protected].

Must Read Articles