Setting Up Policies the Arkivio Way

While many products today talk a good game about policy-based data management, Arkivio is once again ahead of the pack.

On several occasions over the past few months, this column has mentioned Mountain View, CA-based Arkivio in passing—usually in connection with disk-to-disk data migration and data protection solutions. Formed in December, 2000 by four industry veterans concerned about data management, the little company continues to be a microcosm of technology innovation that leads the big development shops in the industry by a couple of years.

The flagship product of the company is software: auto-stor. Auto-stor is described as “a new class of enterprise storage software that manages electronic data and information over its lifecycle for Consolidation, Disaster Recovery Optimization, Data Retention Management, Online Archiving, and Capacity Management of a tiered storage infrastructure.” That’s quite a mouthful to say, and until recently, quite a challenge to get folks to listen to long enough to understand.

Now, everyone is saying it, so Arkivio’s job is getting easier … and more difficult. They have gone from being the innovative stealth company with strange ideas to an also-ran peddling their wares in the shadows of brand name vendors who now claim to have invented what Arkivio has been offering all along.

Still, the smiles are ever-present on the face of marketing communications manager Jerry Huang whenever we run across each other at trade shows and conferences. Jerry has told me over and over again that if I would just take time to look at the product, “to play with it,” I would be all smiles, too.

So, a couple of weeks ago, I did. And, Jerry, you were right. Last month, the company released version 3.0 of its product, which adds more automation as well as move, copy, and delete functionality, and a truly intuitive policy-making engine. While many products I look at today talk a good game about policy-based data management, Arkivio is once again ahead of the pack.

Mercer Rowe, Senior Manager of Field Engineering and Technical Services, took me on a tour of the new release, which was installed live on a Windows 2000 server in his test and development laboratory, and answered all of my questions with candor. He told me about a customer (a large insurance company) who had deployed inexpensive ATA arrays in order to save money on their Big Iron storage investments. They had correctly perceived the inefficacy of storing infrequently accessed data on their most expensive gear and went to an ATA array strategy to provide a second tier of data hosting platforms. However, after cobbling the ATA arrays into their Fibre Channel fabric, they discovered that they had no ready mechanism for migrating the data over to the new platforms.

Enter Arkivio auto-stor. The company deployed the Arkivio central server on a Windows Server, and it promptly went out and discovered the file systems associated with multiple hosts in the environment designated for management by the customer. The product can be used to manage and migrate files stored in an extensive number of operating system-based file systems, including Windows, Sun Solaris, RedHat Linux, IBM-AIX, HP-UX, and various flavors of thin operating systems presented by leading network-attached storage (NAS) vendors. Once installed, agents (about 60 MB of code) are pushed out to the servers that you want to manage, or you can do a discovery operation, produce a detailed report on capacity usage, and decide which servers to manage yourself.

Using an extraordinarily simple interface, the insurance company was able to identify the target ATA platforms for the move and launch the migration. Rowe says the client performed the migration in stages to see what incremental improvement the data re-platforming would provide. First, they migrated all files over three years old, then two years, then one year, then six months. Satisfied with the results, they set a threshold of six months for the automatic migration of files to the less-expensive gear.

The story, while interesting, would not impress were it not for the secret-sauce agents used by Arkivio to expedite data transfers and retrievals. In Rowe’s words, “Our agents strip off the protocol overhead in the Common Internet File System and just send the files at a rate of about 5.5 MB per second on a 100 Mb LAN interconnect.” He added that six-month migration thresholds seem standard in most shops today, for anyone looking for a rule-of-thumb.

This good migration story got better when Rowe showed me the new features of the product. Version 3.0 gets us closer to data management: you can select files for movement based on access date or last modified date, facilitating (to a limited extent) access-frequency-based data migration. Moreover, tools can designate file groups and be used to assign policies to specific groups.

In the latest Arkivio auto-stor, move actually means move. Unlike migration, which follows the old Hierarchical Storage Management approach of moving the data but leaving a placeholder or stub behind, Arkivio's move actually recombines stubs and migrated data and moves the whole kit and caboodle to wherever you want it to go.

Rowe said that the idea of data management tends to garner interest once organizations reach a terabyte or two of storage and about 20 servers. “That’s the point we’ve found when the pain of management becomes less than the pain of buying more disk.”

Another nice feature of the product, added in the latest version, is the policy simulator, which lets you see what impact a policy will have on resources before implementing it. Wedding this to platform cost data could provide the necessary information to justify acquisitions based on reduced cost of ownership.

The limitations of the product are obvious, the key one being a lack of support for databases. This, says Rowe, is being addressed in the next release.

Bottom line: Arkivio auto-stor is definitely worth a look.

About the Author

Jon William Toigo is chairman of The Data Management Institute, the CEO of data management consulting and research firm Toigo Partners International, as well as a contributing editor to Enterprise Systems and its Storage Strategies columnist. Mr. Toigo is the author of 14 books, including Disaster Recovery Planning, 3rd Edition, and The Holy Grail of Network Storage Management, both from Prentice Hall.