We recently sponsored a meeting of a small group of library developers to advise us on our plans for WorldCat Grid Services and an associated Developer Network. While we had a group of geeks in town we thought they would like to tour the OCLC computing and monitoring facilities. I thought at least some of our readers would enjoy this insider’s view as much as I did, so here is my attempt to recreate the experience in a blog post.
When you enter the room that holds the OCLC computing systems the initial reaction is surprise at all of the empty space. In the days of yore hardware took up a great deal more room than it does now, and this is evident in the wide open spaces of this climate-controlled, raised-floor room. And then you find out that this simply one of them, and that there are other rooms such as this on other floors. Most of that space is either rented out or will be rented out to other organizations.
Bill Rogge, who is giving the tour, points out a long structure taller than us that holds a tape library for mass storage. And he means mass. How does ten Petabytes sound? Think you could store enough music and videos on that to keep your teenagers happy? For hard disk storage there is a relatively paltry 180 Terabytes, which is nonetheless still enough to keep your teenagers happy.
We turn around and walk over to an unassuming set of tall, dark metal structures and Bill tells us it is WorldCat (see photo with Jeremy Frumkin of Oregon State University standing next to it). As the group walks on to the next sight I sidle over and touch it. There’s something about touching over 170 million bibliographic records that sends a chill up this librarian’s spine.
The tour continues with a look at the next-generation infrastructure supporting Open WorldCat. Taking a lesson from Google, this set of processors running in parallel is capable of providing millisecond response times by dividing up the index across multiple nodes and merging the results. Using commodity hardware from Dell and open source system software (SUSE Linux), OCLC is working at reducing infrastructure costs while maximizing performance.
All of this hardware is cooled by many sets of high-throughput fans and the entire complex is supported by an uninterruptable power supply. Batteries provide immediate power support for about 45 minutes, which provides enough time to start up and condition power from a set of diesel generators. From there they can keep the operation running as long as they have diesel fuel.
From here we move on to the computer room monitoring facility, which is on the top floor of the building with a view over the highway snaking past the Kilgour building. Projectors mounted on the ceiling fill two wall screens with an array of monitoring processes. Below them is a long table with monitors lined up that display additional processes. This room is staffed 24 hours a day, seven days a week, although today it is nearly empty. It turns out that we arrived on a day when the staff is testing a remote monitoring facility in case they must abandon the Kilgour building. It seems like they’ve thought of everything.
The coup de grace is delivered when Bill saunters over to a large monitor with a joystick and shows how the computer room is monitored by sets of remotely controlled cameras. Taking the joystick, he pans the camera around the computer room we were just in, and zooms in on a fire extinguisher on a pillar, showing how the resolution is such that you can read the text on the tag.
We leave the room, suitably humbled that our bibliographic data is stored, backed up, monitored, and watched with a thoroughness that is almost chilling to behold.
Roy Tennant works on projects related to improving the technological infrastructure of libraries, museums, and archives.