I work on a very hard experiment. Most experiments are hard, even outside of particle physics – it has been a very long time since anyone could simply drop two objects and learn something new. In my experiment we have to worry about the number of Uranium atoms in a few tons of copper. We had to find ways to make electrical contacts and carry signals without the normal mechanisms of solder or wires. We have a cleanroom that we shipped from Stanford, CA to Carlsbad, NM and then put 2000 ft underground in a salt mine. My collaborators are researching ways to grab a single barium ion and identify it with lasers. My experiment has overcome a lot of obstacles, but in some ways the most recent problem is the most bizarre: how do you get the data out of the salt mine?
Particle physics experiments usually result in huge amounts of data, which always pose a challenge to process and store. The LHC experiments see Terabytes (1000 GB) of raw data a second, which is filtered and recorded at around 1.8 GB a second. Our maximum rate is around 80 MB/s (about 4% the LHC rate), which we will sustain for a few hours during our calibration periods. The big collider experiments have way more data, but they also are within state of the art computing facilities. My home network can’t handle our data rate, but the almost any academic network could. If we were trying to get this data from one side of SLAC to another, it wouldn’t be much of a problem. However, we are trying to get it out of a government salt mine.
WIPP Entrance (note the fence)
The WIPP site has incredible security, which I’m sure everyone appreciates since it is a nuclear waste storage site. However, their security protocols mean we can’t use their network. We have our own 1.5 MB/s network going in to the mine. At least the last time I was there (back in October), the network wasn’t always working. It will never be good enough for transferring our data. It is barely enough to transmit commands to control the experiment remotely, such as when the mine is closed. We looked into trying to get a better connection, but that would mean getting a new line installed between WIPP and Carlsbad, which is something like 35 miles. It might be cheaper to get our data to the moon!
Since the data can’t get out via wires, we are left with a choice between tapes and hard drives. Tape drives might seem old-fashioned, but they are still used in science. They are less fragile than hard drives and giant tape-loading robots simplify the data transfer process. The fragility is an issue since we will be shipping the drives back and forth between WIPP and SLAC. Our collaboration has found only one experiment so far (SAUND, in the Bahamas) that ships hard drives. Their data volume is also less than ours – we’ll be practically juggling disks going back and forth. We’ll have the delays of transferring the data from the WIPP computers to the drive to be shipped (or at least backing it up), the transit time, and then the time of getting it off the transfer disk.
So forget determining the neutrino mass – that’s easy. How do we get our data from WIPP to SLAC with the minimum delay and cost? Moving those bits, with all of our physics in them – that’s hard.
