"We're hoping that using Caringo's CAS will replace 700 GB of our network attached storage (NAS) file shares and drive our backups down from 3 terabytes (TB) dailies and 4 TB fulls to 1 TB dailies and 2 TB fulls," said Chris Schremser, chief technology officer (CTO) of ZirMed.
While many in the healthcare field originally ran across CAS for HIPAA archiving, Schremser said it was tape backup pain that drove him toward it. "I can't tell you how sick I am of tapes, tape backups and tape backup software," Schremser said. "I didn't want to deal with any of that so much anymore, but I also didn't want to have to juggle file shares."
Each insurance provider requires his company to break claims down into smaller transactions and send them as single files, Schremser said, which would make managing it all on conventional NAS a nightmare, especially since so much of it, some 300,000 of those small files per day, is unchanging content.
Schremser said he evaluated EMC's Centera product first. "We looked at it very seriously …We had sales engineers in several times and calls with API programmers about our specific use case." However, they never brought gear in-house for testing.
Xiotech referred ZirMed to Caringo, a new "open" CAS company co-founded by Paul Carpentier, who sold FilePool to EMC in 2001 for a little under $50 million. FilePool's product became EMC's Centera.
As CTO of Caringo, however, Carpentier is arguing that Centera is no longer the best implementation of CAS. Caringo's product, CAStor, has three notable features that differentiate it from Centera: It provides a patent-pending upgradable hashing scheme, an HTTP interface and is hardware agnostic. Caringo's current claim to fame is the tiny thumb drive on which its software is delivered. The company has been seen demonstrating its installation on Windows laptops and Apple desktop drives at industry shows.
"Centera was impressive," said Schremser, pointing out that Centera includes offsite replication while Caringo's system still only does local replication. In fact, Schremser said his planned move to put the Caringo system in production will have to wait until late May, when a beta version of wide-area replication software will be available.
But, he said, the deal breaker with Centera was the price. "They kept talking about it being built on commodity hardware, but then their price was well over six figures," Schremser said, declining to specify the exact price. It was such an insurmountable cost for his company that for a while it even contemplated hacking a homegrown CAS system using a database management program.
By contrast, ZirMed is now paying a total of $20,000 for two 6 TB Caringo systems, one for the company's headquarters and another for a remote disaster recovery site. The systems each include a cluster of three 1U "pizza box" servers for storage and Caringo's software, for which Schremser said the fee will be approximately $5,000 (as long as the software is in trial, the company is using it for free, Schremser said).
In order to hook CAStor into its production environment, ZirMed's homegrown production applications have had to be rewritten to "point" at the Caringo product and will need to be rewritten again if the company decides to get rid of it. "It's like any Web download … there is still an importation process for data whenever you add it to a new system and a process to remove it," Schremser said.
However, as homegrown applications, Schremser pointed out they would also have had to be rewritten to access Centera on both the application and the Centera side, "and that would have been another significant sum." Caringo, meanwhile, speaks HTTP protocol, which means there is no change needed to allow it to receive ZirMed's data.
Schremser said he's also aware that Caringo doesn't have the kind of support EMC does for applications like Microsoft Exchange. "That's not what I needed it for … It's more important to us that this integrate with our workflow, keep our data safe but online and retain information about the lifecycle of the data," he said.
The proof, of course, is still in the pudding. "There is some risk; I know I'm an early adopter," Schremser said, adding that he was going to continue the dreaded tape backup cycle for several months even after CAStor goes into production as insurance. The product is also undergoing heavy testing in his shop, but so far, "I have no reason to doubt at this point that they're not capable of delivering what they've promised," he said.
Reached for comment, Carpentier said wide-area replication "will be one of the defining features of [version] 2.0" of Caringo's software, due out in late May. The replication Caringo will offer, Carpentier claimed, will include capabilities like the replication of subsets of data to multiple remote clusters of CAStor.
EMC emailed SearchStorage.com the following response: "The world has many options for information archiving, and organizations make the choices they view as best for their needs. EMC Centera can scale from 2.2 TB of useable capacity to petabytes and has addressed the archiving requirements of small and large organizations alike. Its features and functions are unmatched."