Dell executives laid out much of their roadmap plans for customers during this week’s Dell Storage Forum 2012. The company is in the second year of a major storage transformation following its break-up with former partner EMC and a new reliance on its own technologies.
“In the past, Dell has been criticized for not talking about the future, where we’re going,” said Carter George, executive director of Dell storage strategy, during the conference’s final keynote on Wednesday. “The relationship with your storage vendor is important. Typically, it’s a long-term relationship. When you go on a journey together, you want to know where we’re going.”
One key piece of the Dell storage products strategy is a flash cache product, code-named Project Hermes. Hermes includes cache coherent PCIe solid-state storage cards in servers, and software that processes reads and writes in memory located in the server. The idea is to move storage closer to computing power, but Project Hermes addresses the challenge associated with siloed direct-attached storage (DAS).
“In three years, all active data will be on flash or some kind of memory,” George said.
Dell calls this technology “Fluid Cache.” It transforms memory as a fixed storage resource within the server and makes it a shared resource among servers; it comes from cloud memory startup RNA Networks, which Dell acquired last year. The Project Hermes product is scheduled for general availability in the first quarter of next year, according to George.
“With RNA, we will put tier-one storage in the server and make it cache coherent across multiple servers,” said Bob Fine, Dell’s director of product marketing. “I’m not saying how we’re going to do it, but the concept is to imagine putting tier-two and [tier-]three storage in the array and tier-one storage in the server.”
Project Hermes is Dell’s response to EMC’s VFCache, according to Randy Kerns, a senior strategist at Evaluator Group. He said it spreads storage resources across servers instead of relying completely on the storage-area network (SAN).
“They’re trying to figure out how to share [storage] resources for multiple servers. You put a cache card in the server,” Kerns said. “You can do load balancing and capacity balancing. It’s all about a shared resource issue.”
Placing a read and write cache in the server could set Dell apart from its competitors. “It’s different than everybody else. As far as I know, nobody else is doing the write cache,” said Terri McClure, a senior analyst at Enterprise Strategy Group (ESG).
Dell customer Paul Traue, a system administrator at MindBrix LP, a North Richland Hills, TX.-based software development company in the mortgage industry, found Fluid Cache “extremely interesting.” But, he was quick to add, “Being a release 1.0 customer would scare me to death because there are a lot of problems to be solved there.”
ESG’s McClure said 80% of the typical operations are read, but there’s a considerable overhead associated with the writes that make up the remaining 20%. She said virtual machine environments “tend to be a little bit more write-heavy than read-heavy because of the way virtual machines handle I/O.” So the read/write cache gives Dell a “little extra added advantage and helps a little bit more in virtual environments,” she noted.
Dell is also working on a placing an object storage layer -- code-named Bob -- within its EqualLogic and Compellent storage arrays for a cloud play. When asked if the object storage will be based on erasure codes, George responded “could be.”
Primary dedupe still on drawing board
Ocarina Network’s data reduction technology will also play a significant role in the Dell storage products line. George, who was Ocarina’s vice president of products before Dell acquired the startup in 2010, said every Dell product will have Ocarina technology baked into it. Dell released a DR4000 disk backup system using Ocarina deduplication in January. Now it’s integrating Ocarina’s technology into its fluid file system. The integration is expected to be code complete in July and scheduled for release in January next year.
An industry insider with knowledge of Dell’s plans said it will include compression in its Compellent and EqualLogic storage platforms first, with deduplication to follow.
Host virtualization, heterogeneous replication for DR
Another Dell high-priority project is called Host Virtualization Storage (HVS). HVS virtualizes the storage workload for read/write processing, said Laz Vekiarides, executive director of storage software engineering at Dell EqualLogic. The product is expected to encapsulate storage as an application, so storage can run as a virtual machine.
“This is a brand new area,” Evaluator Group’s Kerns said. “It’s about running storage as a component.”
MindBrix’s Traue said there’s a high probability his company will use HVS. “It would make [disaster recovery] DR really, really nice,” he said. “Replication is the bulk of my DR strategy, and being able to pull some of the physical arrays that I’ve got out at my DR facilities as primary storage and wrap that into hosted virtual storage could save me a lot of money.”
Dell also has big plans for backup with the AppAssure data protection software acquisition it made this year. The vendor will develop AppAssure technology for heterogeneous replication and snapshots. The function is expected to be embedded in its storage arrays, so a Compellent array can replicate to an EqualLogic array and vice versa.
In addition, AppAssure will enable the ability to recover from a virtual machine if a physical machine crashes, even if data wasn’t backed up to a virtual machine. “You can have full disaster recovery, but only pay for it in the event of a disaster,” Dell’s George said.