Comment Xiotech wants CorteX to be the door to a new storage world. CorteX is a REST-using API that lets an application access and manage Xiotech's storage brick, its ISE (Intelligent Stage Element), the sort of super disk drive with fail-in-place components, a five-year warranty and terrific performance from its multiple spindles.
Steve Sicola, Xiotech's chief technology officer (CTO) enlightened us about the ISE, its background, and where it is headed.
Xiotech gained the ISE technology when it bought Seagate's Advanced Storage Architecture's group and promptly developed its Emprise storage array based around it. This was a successor to the then-extant Magnitude 3D, a classic dual-controller, modular array, a me-too array that has now disappeared. Xiotech has stopped focussing on small-to-medium enterprises and has been aiming its marketing a products at larger businesses for the last four months.
Xiotech expect people to buy ISEs as needed, aggregating them together into a stack of bricks which, together and individually, store, move and protect data. Each ISE is a self-healing, closed box with no field-replaceable unit (FRU) components. Sicola says: "ISE is a satellite; it's not coming home."
ISEs don't carry all the upper storage stack data management software that a dual-controller and modular array does - say a CLARiiON or an EVA. The job of storage data management should be carried out by applications running in servers. In Sicola's view: "VMware has strong thin provisioning, so array controllers don't need it."
Dell, EMC, Fujitsu, HDS, HP, IBM, LSI, NetApp, NEC, Promise and every other manufacturer of the classic dual controller, modular array or monolithic array have got it wrong. Server virtualisation has changed their world and the storage data management role is transferring to hypervisors and software like Symantec's FileStore. They say that if you provision, replicate, snapshot and migrate storage then the brick works faster and does what it is told better.
Sicola says: "ISE is meant to be told what to do by the upper levels of the storage stack," and CorteX is the way it should be done.
There are three parts to the Xiotech pitch. First here is the ISE itself, then the ICON Manager for it, which can run as a virtual machine and will plug into VSphere, and thirdly the CorteX API set. There are six individual APIS in Cortex using REST, and it only took two months to create.
CorteX is about making the ISE integratable everywhere. "It is a RESTFul service running directly on the ISE, and means that ISE can be treated as a software component in the same way as other [software] components in a virtualised infrastructure. Resources are identified by URIs, which can be used to obtain information as well as to alter system defaults and manage volumes."
Today an ISE NAS is coming, a NAS head running on a server which uses ISE as its storage. It's a joint announcement with Symantec and the NAS head is Symantec's FileStore. Huawei is also associated with FileStore, and Xiotech executive staff have visited China. We wonder if there is scope for an OEM agreement between Huawei and Xiotech. Sicola won't say anything about any OEM deals other than: "We're talking to everybody."
He points out: "We're supported by [HDS'] USP-V, [IBM's] SVC and FalconStor. You can bet there will be a NetApp thing. We can make NetApp fly." Is he alluding to NetApp's V-Series working with ISE here? He wouldn't say.
Sicola says ISE is much more reliable than the equivalent number of hard drives inside a traditional storage array: "We're a quarter of the code of an EVA or CLARiiON, and half of our code is error-recovery, self-healing stuff. One customer I know replaces 10,000 enterprise hard drives a year. It's crazy. With ISE rebuilds I can read most of the stuff on a failed drive. Other supplier's error recovery processes can't do that."
Xiotech has built around 3000 ISEs and shipped 2,500 of them. The company constantly monitors them, getting telemetry back, and this data is used to improve the reliability of ISE products. An ISE is self-healing. There is no need with it to do RAID builds anymore. Sicola asks us if we remember the Black Knight from Monty Python: "If anything in ISE fails, it's just a flesh wound."
Where did the name ISE come from? "The term was coined by Apple marketing in 2006 because they didn't like the brick word. Seagate contracted Apple."
SSD and ISE
What about solid state drives (SSD) and ISE? Sicola said: "We're going to be supporting SSD in the future. SLC (single-level cell) is the way to go. MLC (multi-level cell) will burn out soon. [With] bigger MLC parts the number of writes go down by half every time. Other technology will come up behind SSD and be more reliable.
"SSD has more ways to fail than hard disk drives. The makers don't know how to do error-recovery. The HDD guys have all the [error-recovery] patents. Seagate will end up just fine in the SSD business.
"Major customers are using SSDs to hold VM read images. When I do SSDs it will be all about I/O density."
Sicola says: "The big deal right now is international expansion." Xiotech is looking to significantly expand its selling operations into new territories such as Asia, Australia and New Zealand.
Sicola says ISE has a much lower total cost of ownership (TCO) than traditional storage arrays and even newer ones such as the Compellent product. He cites a three-year TCO study which showed a Compellent installation costing $476,000 versus a ISE equivalent costing $53,000. He says a 3U ISE box can support 11,000 Ecxhange users. Figures like these are being used to recruit channel partners and help them sell into enterprises who can do more and buy less.
Xiotech is impatient. Its new management has made the ISE team the central focus. Storage veterans like Richie Lairy have been persuaded by the new team to come out of retirement and get to work in the ISE sandbox. Xiotech is the largest privately-owned storage company out there and you get the distinct feeling it wants to lose that tag by having an IPO as soon as it can. That's dependent on making ISE a success, and that means CorteX has to deliver the new storage world of dumb, high-performance and very reliable storage bricks doing their thing in virtualised storage infrastructures at the behest of hypervisors and other storage-using applications. ®