Baby steps for open source storage

Continuing on the topic of open source storage from last week, I would like to wish a belated happy birthday to the Aperi project, the first anniversary of which passed last month. I probably was not the only one to miss marking the occasion, as its first public update on the project went generally unnoticed.

[ Mario Apicella's column is now a blog! Get the latest storage news from the Storage Adviser blog. ]

Much discussed when launched a year ago, this comprehensive, open source storage management framework has since divided vendors into two opposing groups: those that actively support the initiative (IBM, for one) and those that could not care less, such as Sun, which, after providing initial support to Aperi, moved out and -- like many other vendors -- has been out ever since.

Could that be why Aperi didn’t receive many accolades on its first birthday? Whatever the reason, it’s hard to justify the inattention because those involved with Aperi have made progress worth noting.

Click for larger view.

For example, take a look at this architectural diagram of Aperi. The server core -- that big rectangle in the center -- rests on an Eclipse foundation and hosts services such as auto-discovery of the storage network, as well as device monitoring and control. That application currently runs on Windows and Linux servers and will probably stay that way going forward.

The bottom part of that chart shows the host agents. As I understand it, these pieces -- obviously important to the storage management puzzle -- could use help from developers skilled in operating systems such as Solaris and HP-UX. As that section of the diagram shows, Aperi’s commitment to the SMI-S protocol has not relented.

At the top of that chart is probably the most critical layer for Aperi: management plug-ins. Think of the device drivers in Windows or any other OS. Obviously, that’s an area where contributions from equipment manufacturers are of paramount importance. After all, should that contribution be missing, IP rights may prevent creating or distributing those plug-ins. Moreover, reverse-engineering to expose the logic of a complex piece of equipment can only go so far. We’ll see how that part turns out as the project evolves more.

The terminal screen that represents the UI on the left side is self-explanatory, but it’s worth noting that Aperi is not limited to the OS Derby database as the icon on the right seems to indicate. Aperi has been tested also with DB2; other database systems, including Oracle, should follow before launch.

Speaking of launch, a date has yet to be determined, but you can download a runtime version of Aperi, which contains only the binaries, install it on a Windows or Linux machine and try it out yourself. Because it is a work in progress, don't expect a lot right out of the gate. For example, the current iteration of Aperi does not support multi-pathing.