The Art of the Archive

Backing up data for a university campus – when the users are a widely dispersed mix of students, faculty, researchers and others – isn’t easy. Archiving that data at a large facility like the University of Western Ontario can be downright intimidating.

Western moved recently from tape storage to a disk-to-disk system to back up 25 TB of information a month. This put an end to several problems, including tape failures and getting tapes locked in drives, and it raised the IT department’s service levels noticeably.

But the job is not complete. According to Jon Hickmott, Western’s senior database analyst, the London, Ont.-based institution has not yet decided how it will archive all of its growing data in the future.

Part of the problem is the number of sources and types of University data, and the fact that much of it is used for very different purposes. And some data is accessed often, some not.

In addition, while the central IT department looks after the administration computers and the servers that deal with the e-mail and the Web demands of most users, department servers have their own little IT groups, some of which run their own e-mail systems and other applications.

With more than 300 departmental servers, almost all with their own local storage, plus other similarly far-flung servers, the central IT department must back up roughly a TB a night over the network to its own local servers, which are then hooked to large disk arrays.

While the disk-to-disk transition has been made, several more changes to Western’s backup environment will be in place by the end of February 2006.

It will soon be composed of two Solaris Servers (v880, Solaris 9) running EMC Legato Networker 7.2.1, with one master server and one storage node; a Falconstor VTL server with two virtual tape libraries (one for each server) and 12 TB of disk on two Nexsan Ataboy2f disk arrays and an IBM DS4100 disk array; plus two StorageTek tape libraries, one onsite and one offsite.

Still, many decisions about archiving have not yet been made, so Hickmott and Dennis Regnier, Western’s associate director of technical support, have gone in search of answers.

“We’re looking for a solution that will help us manage the backup and let us clean up our high-priced production disk,” says Hickmott.

“But being in a university environment, we have a number of issues around all the different types of data we have. We’ve got research data, and then we have financial data as well.”
While such data does not have to be kept separately, when it comes to archiving, it is difficult to put policies on it, especially on how people should be allowed to move it around or how it is deleted in the end. A lot of the data is owned directly by students and faculty.

“For us to come in and say, ‘OK, after 30 days if you haven’t touched that data it’s going to be moved off to a tape somewhere,’ might not work for them because two months down the road they might need all that data back and can’t handle bringing it back to disk at that time,” Hickmott says.

Meanwhile, the data keeps expanding. Hickmott expects about 25 per cent growth this year.

So Western is considering various archiving solutions, including MAID (massive array of idle disks) arrays which, in an archiving environment, take data then spin down the disks to save power and the disks’ operating lives.

Also, while Hickmott’s department offers a backup solution for the entire campus, it doesn’t really offer a centralized disk on which users can “take a chunk of space,” besides a central Windows domain with a cluster that some smaller departments use.

But the department wants to offer such a service. It has taken a serious look at i-SCSI, since bandwidth is abundant on campus and the technology is quite straightforward to use.

“It’s a cheap solution for them to slap a network card in and voila, they have a block of disk that’s (accessible) off a central area in our machine room.”

Besides, with a number of researchers on campus, central IT is trying to find a solution that is economical and fits their funding model.

Being grant-funded, many groups tend to receive a lump sum at the start and must buy technology right away, then live with it no matter what for the rest of the grant period. Being able to provide such groups with a block of disk and an upfront cost would be very useful, Hickmott says.

In the end, Western’s archiving solution must take into consideration the reality that students must be able to access course data whether they are at home or somewhere on campus with a laptop or a PDA.

So the research by central IT goes on. Hickmott and Regnier continue to check out technologies and attend conferences, including the Ontario Universities Computing Conference (OUCC) going on this year in May at the University of Guelph.

Due to the complexity of the technology and the challenges of a campus environment, technology decisions cannot be taken lightly at Western. As such, they can take some time to get right, argues Hickmott.

“We have to provide services that the students and the faculty can take advantage of to help them with their data needs. It’s not just a typical finance department and a few databases.”

Would you recommend this article?


Thanks for taking the time to let us know what you think of this article!
We'd love to hear your opinion about this or any other story you read in our publication.

Jim Love, Chief Content Officer, IT World Canada

Featured Download

Related Tech News

Get ITBusiness Delivered

Our experienced team of journalists brings you engaging content targeted to IT professionals and line-of-business executives delivered directly to your inbox.