Hi all,

Just to follow up on Ron's comment, which he and I discussed on the way out of the TechAssess telecon this afternoon...

On Jul 31, 2006, at 4:07 PM, Ronald Lambert wrote:

My feeling is that current disk MTBF is much less than that figure, and would propose writing a minimum of two sets on the summit to maintain data integrity. If we have to drive disks down the mountain in case of fibre failure we don't want that to be the only copy.
ron


I think a good argument can be made for two copies of the data on the mountain, based on at least two points...

- we discussed a "policy" that there should ALWAYS be two copies of the data on spinning disk. To implement this as soon as possible, one needs two copies of the data on the mountain. This is particularly true if the mountain-base link is down, and the base copy doesn't get written.

- this is the case Ron mentioned: if the mountain-base link is down for more than (say) 24hrs (not gonna happen except in really bad circumstances), then our fallback position is to unplug a set of disks (probably 1 4U unit) and truck it down the mountain. One CERTAINLY does not want to throw our ONLY copy of last night's data in the back of a truck that will be traveling down a dusty, curvy road. One wants a second copy to stay safe and sound on the mountain until the traveling copy gets written into the base architecture and transferred to the archive center.

I'd imagine that this is not a significant cost factor... only a few TBs. :-)

        Cheers,
          Chris

======================================================================== =
 Dr. R. Chris Smith                    EMAIL: [EMAIL PROTECTED]
 National Optical Astronomy Observatory
 Cerro Tololo Inter-American Observatory
 950 N. Cherry Ave.                    Casilla 603
 Tucson, AZ 85719                      La Serena, CHILE
 Office: 520-318-8555                  FAX: 520-318-8170
 WWW: http://www.ctio.noao.edu/~chris
======================================================================== =


_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data

Reply via email to