Reporter's Notebook: Whispers and rumbles from SNW
Floridians talked DR, big guys quibbled over deduplication, and rumors of acquisitions and new products to come flew at SNW Orlando this year.
In the wake of Wednesday morning's announcement that EMC Corp. is to acquire super deduper Avamar Technologies, Hitachi Data Systems Inc.'s (HDS) chief technology officer (CTO), Hu Yoshida, explained why his company is sticking with dedupe within its rebranding of Diligent Technologies Corp.'s ProtectTier VTL.
While Avamar uses agents on each host to dedupe before data is sent across the wire, Yoshida said he prefers the Diligent approach, which dedupes in a central repository at the back end, keeps a 4 GB index in a central memory repository and can slide into the backup environment.
The centralized approach, Yoshida said, "means you can look at the whole spectrum of the data and keep it all together in that big RAM [random access memory] repository."
Meanwhile, Avamar's founder, Jedediah Yueh, pointed out that pulling data across the wire and deduping it all in one place wasn't necessarily the most efficient from the Avamar (and EMC) perspective. "We split the index over multiple servers; they pull it all across the wire and can sometimes choke on it," he said.
But still, experts said, the buzz picked up around dedupe might mean the EMC acquisition has touched off another buying fad among big companies, with dedupe companies that include VTL supplier FalconStor Software and nearline storage vendor Data Domain Inc., as yet unclaimed.
Might HDS be looking to acquire Diligent? "No comment," Yoshida said.
Gossip, gossip, gossip: Whispers and rumors
Microsoft was dropping hints that another big company will come out with a new rebranding of its Windows Storage Server 2003 in the next few weeks -- and according to the SNW buzz the most likely candidate is IBM, which OEMs iSCSI/NAS products from Adaptec Inc., which in turn just released a new version of its iSCSI target software that includes automated provisioning and Volume Shadow Copy (VSS)-based snapshots. According to sources, the product could be positioned as a competitor to Hewlett-Packard Co.'s (HP) All-in-One (AiO) box, announced in early September.
Word also has it that Onaro Inc. was on the block for acquisition and received offers from two of its major OEMs, but it turned them down as they came in too low. The scuttlebutt around SNW is that Onaro "may have missed its boat" by turning those offers down, according to one source.
"Sometimes companies miss the best offer, and then they stand a bigger chance of ending up on the 'where are they now?' list," said an analyst, requesting anonymity. Onaro declined to comment.
Other sources reported that NEC Corp. had been briefing analysts on its plans to release a new VTL. "It'll be what VSM [Virtual Storage Manager] Open should have been," said one industry expert, referring to Sun Microsystems Inc./StorageTek's failed efforts to bring out a high-end open systems VTL.
Speaking of Sun, a persistent little bug was making the rounds at the show that vice president of storage marketing and business operations Nigel Dessau is soon to be leaving the company, a notion categorically denied by Sun representatives contacted by SearchStorage.com.
End users mull over interoperability at SNIA session
During a session on Wednesday afternoon, the Storage Networking Industry Association (SNIA) End User Council presented preliminary results of its survey of more than 400 end users over the past year. SNIA doesn't yet have all the hard numbers of the survey worked out, but the early results indicate that interoperability issues between different vendors' products is a growing frustration among end users -- a contention borne out by a lively discussion that ensued among attendees, some of which also attacked the SMI-S standard that is the SNIA's answer to the interoperability woes.
"SMI-S doesn't help me -- it's not being written fast enough," said an attendee from a large East Coast financial company who asked that neither he nor his company be named for legal reasons. "The support matrix certainly doesn't help either. I'm wondering if there's a simpler answer -- if all the vendors could standardize on just one database that would indicate at the end-user site if firmware or software versions were due for upgrades within the environment."
Another user was even more direct about what could be done to make vendors interoperate better: "Torches and pitchforks," he said.
Floridians talk disaster recovery (DR)
Disaster recovery is a hot topic in hurricane alley -- last year's SNW went on while Hurricane Wilma raged across the southern portion of the state, flights were grounded and, for some, all hell broke loose in the data center. This year, users talked about how they're better prepared.
"Wilma took out Naples, then moved further north and took out Fort Lauderdale, West Palm beach and Port St. Lucie," said Ben Weinberger, IT director for Florida law firm Ruden McClosky, in a presentation Thursday afternoon. Tampa, St. Petersburg and Sarasota are very close to Naples and were lucky to also escape disaster during Wilma, he said.
"There's really no place to hide in Florida," Weinberger said.
However, after arriving at the law firm in September 2004, Weinberg had set about fortifying the business to weather the storm. Weinberg and his staff centralized data at two offices, one in Tampa and one in Fort Lauderdale. The company uses a combination of XOsoft Inc./CA's WANSyncHA and Riverbed Technology Inc.'s Steelhead wide area network (WAN) optimization appliance to replicate data from 10 branch offices back to these central sites.
By sharing the location at the other end of the wire and reusing hardware wherever it can, Weinberg said, "We got all of this done for less than $100,000."
With around 15 terabytes (TB) of data, Ruden McClosky is one example of many small and midsized shops putting together big-time disaster recovery because of its location in the sunshine state. Most shops either have or will be implementing long-distance failover and replication similar to Ruden McClosky's and using cooperation or managed services where it could to save money.
"We're looking at sending our data to a collocation with Georgia Tech [Georgia Institute of Technology]," said Rea Burleson, director, university computer systems with the University of South Florida, referring to a practice common among cash-strapped colleges and universities in implementing disaster recovery. (See Universities lean on each other for better disaster recovery, Sept. 7).
The Florida Turnpike Authority, too, is looking at failover, synchronous replication and geographically dispersed clusters between its two data centers at headquarters in Boca Raton and what has, until now, been a secondary site in Orlando. According to its storage engineer, Ed Richard, in order to get its approximately 10 TB of toll information zinging across the data centers, the turnpike is putting in three 10 Gbps fiber lines under hundreds of miles of road.
