Tech Assess concall - June 19,2006

Attending : Don (LLNL), Michelle and Chris (NCSA)

We discussed several alternatives for recovering from a broken link
between the summit and base.  There will be sufficient disk at the
summit to buffer 5 nights of data.  There are 3 alternatives for getting the data down to LaSerena.

1. Tapes - The telescope operator can manually mount the tapes and
arrange for them to be transported by car to the base camp.  There
will be a small number of tapes - perhaps as few as 3.  This requires
additional new hardware at both the summit and base with additional
costs for spare parts and operator training.

2. Physical transport of disks - Drawers of disks could be dismounted
and pulled from the storage rack at the telescope and driven down to
LaSerena.  Repeated moving of the disks can only increase the failure
rates.  To achieve the required disk write bandwidth, we will probably
need to create stripe sets that go across multiple disk drawers, so
physical movement of the disks will require moving *all* of the disks
and reconnecting them correctly at the base.  We consider this to be very risky.

3. Networking - Once the summit/base network is fixed, we can use the relatively idle network bandwidth during daylight to begin the
transfer of the data to the base camp.  Five nights of data can be
sent to the base in 5 days. We will need the 10 Gb/s link back to the
archive site (that Jeff has hinted at several times) since the 4 Gb/s
bandwidth I have asked for assumes a continuous 24 hour transfer time of nightly data under normal conditions. This solution adds no hardware and no additional costs.

Next concall is Monday, June 26 at 10AM Pacific time.