Nimble Storage Thursday came out of stealth with a storage system that the startup’s executives said combines primary storage with deduplication for backup in the same device. It makes sense that Nimble would use dedupe, considering its founders were former Data Domain engineers.
But Frank Slootman, president of EMC’s data backup and recovery division and Data Domain’s CEO until EMC acquired the company last year, says there is no dedupe in Nimble’s storage. Slootman saw my story on SearchStorage about Nimble, and sent an email claiming “there is no dedupe in
Nimble whatsoever. Read their white paper, or just ask them. We did. They do have local compression.”
I did ask Nimble CEO Varun Mehta when I spoke to him before their launch. He said his storage systems use inline compression for primary data and dedupe for backups. And according to Nimble’s press release on its product launch (emphasis added):
The CS-Series is based on the company’s patent-pending architecture, Cache Accelerated Sequential Layout (CASL™), which enables fast inline data compression, intelligent data optimization leveraging flash memory and high-capacity disk, instant deduped backups, and WAN efficient replication – all in a single device. CASL allows organizations to reduce their capital expenditures for storage and backup by at least 60 percent, while eliminating the need for separate, disk-based backup.
And a data sheet on the Nimble web site states:
Nimble slashes IT costs by converging compressed primary storage, deduped backup storage, and disaster recovery into one solution.
Slootman is correct about the whitepaper, though. A paper called “A New Approach to Storage and Backup” on the Nimble site does not say it uses deduplication. It claims “Nimble Storage CASL provides in-line compression on all data” and in a section on its backup technology says “CASL enables instant, application-consistent backups on the same array with very efficient (up to 20x) backup capacity optimization.”
Capacity optimization could be dedupe or compression. But nowhere in the 15-page whitepaper does Nimble claim to dedupe backup data.
While Nimble execs said in press interviews that they dedupe, they had a different message at a blogger TechField Day in Seattle where the startup officially launched Thursday. Nimble presenters did not mention deduplication at the blogger event.
I asked Nimble for clarification about its mixed marketing, and its VP of marketing Dan Leary replied via email:
“Sorry if there was any confusion regarding deduplication. Nimble does not deduplicate in the Data Domain sense, where all duplicate blocks are eliminated using a content-based signature. Our snapshot-based block sharing eliminates duplicate blocks across backups like deduplication systems. Nimble compresses, but does not deduplicate, within a primary storage volume. However, we offer better space savings compared with any secondary storage. Secondary storage systems require a baseline copy of the original data to get started. Because converged storage doesn’t require a baseline full backup, Nimble provides even better capacity optimization than secondary storage. Look for an upcoming blog from our CTO who will cover this topic in more detail.”
If Nimble can shrink data enough to make backups and replication for DR more efficient without taking much of a performance while compressing, it may not make much of a difference how it’s doing it. Nimble beta tester Dave Conde, IT director of eMeter, says he’s found performance outstanding and he’s getting a reduction in data although he hasn’t measured the actual rate.
But if Nimble is deduping, EMC execs probably want to know just how close the startup’s dedupe technology is to the dedupe it paid $2.1 billion for when it acquired Data Domain.
In a follow-up email, Slootman attributed Nimble’s mixed message to “a disconnect with marketing. They probably mean like NetApp that their snapshots use block differentials. They should not be using the term [deduplication] so indiscriminately.”