A research facility in Saskatoon is using Fibre Channel and ATA networked storage devices to back up and archive terabytes of data for from their business operations and from experiments.
CLS, owned by the University of Saskatchewan, houses a 2.9 Giga electron volt synchrotron, which uses radiofrequency waves and magnets to accelerate electrons, causing them to give off extremely bright light.
The light can be sent down a path called a beamline, and used to get information on various materials. Applications include biomedical imaging (from which researchers could glean early cancer detection methods) and microcharacterization of materials (which can help researchers learn about the effects of anti-wear coatings in automobile engines).
All of these experiments create terabytes of data, said Christopher Angel, CLS’s IT systems analyst.
For example, he said the biomedical beamline is not even live and already needs a terabyte of storage, while the protein crystallography beamline requires 3TB and can create 70MB of data ever 10 seconds.
“Biomedical eats data like it’s candy,” Angel said.
CLS currently has two EMC CLARiiON disk arrays – a 500GB ATA and a 300GB Fibre Channel model – and the shelves installed this week are intended to replace an ADIC Scalar 1000 LTO2 tape library.
“We saw that we were filling up a library with tapes that were just going to sit there for two years,” he said. “We were outgrowing that library. It was going to be ridiculously expensive for us to get a new library, so we looked at alternatives, and we decided to pick up a 500GB shelf, which gives us about five and a half terabytes of usable storage, and we devoted that to take off some of the higher rotation storage out of the library.”
He said this will be used primarily for 30-day backups of business data, so he can recover files or folders deleted by users more quickly – in some cases, cutting recovery times from 15 to five minutes.
“If it’s, ‘Please recover this entire directory that we’ve been working on for 10 days, from nine days back,’ and the last full backup was 20 days ago, instead of digging through all these tapes, I only have to look at a disk and one tape, so it’s much much quicker.”
Experimental data will be stored differently due to a legal obligation to prevent clients from looking at other clients’ data.
“In some cases, they would prefer that we don’t keep their data,” he said.
“If you have some guy coming in, and he’s got his mult-billiion dollar experiment, and he doesn’t want his competitor, who happens to be using the beamline across the building, stealing his data.”
The EMC storage devices meet both Fibre Channel and ATA standards, said Michael Kerr, EMC’s director of Canada channels.
“Over time, you can move the data from the Fibre Channel fabric to the ATA fabric,” Kerr said. “The unique capability of the CLARiion to house both Fibre Channel and ATA together is what makes this very cost-effective. So rather than having to tier it to a separate server, it’s all done within the same physical frame.”
But CLS is not using Fibre Channel to connect the storage devices directly to the computers used to collect experimental data, although that was the original plan.
“Instead of getting a 2 Gbps Fibre Çhannel connection, they’re just going over gigabit ethernet and that’s providing them with more than enough throughput,” Angel said. “In the future if they need it we have the capacity
to plug them directly into the (storage-area network) but right now we don’t really need it.”
Comment: [email protected]