Questions and Answers (09/14/2006)
Q1: Will utilization increase?
Answer: Yes, some growth is possible.
Q2: What is your target number for availability? 97% (which NCCS reported) seems low.
Answer: Although NCCS can't get 100%, NCCS will put system back ends and mass storage onto UPS to increase availability.
Q3: Will NCCS show the wait time statistics in the User Forum?
Answer: NCCS should incorporate queue wait time info in the next User Forum statistics.
Q4: When the cluster becomes available, how will testing of upgraades be done?
Answer: A subsystem of 32 nodes for testing will be orderd. For now, NCCS will carve out 16 nodes for testing.
Q5: Is another organization building a cluster like NCCS's?
Answer: Yes, ARL is building one, and we have started a learning relationship with ARL and the vendors. We will take a multi-pronged approach to vendor relations.
Q6: Although the commodity architecture is cheaper than alternatives, must we do costly in-house work to support it?
Answer: Yes, but in the TCO we included costs for work on our side.
Q7: Will the cluster have UPS?
Answer: Yes.
Q8: Is the schedule change (delaying cluster operation) affecting the schedule for 'halem' decommisioning?
Answer: Halem will remain available until which time we need cooling capacity for the new system.
Q9: Is there a direct route between Columbia's front end and 'palm'?
Answer: Yes.
Q10: Can user use bbftp to get to 'lou' at NAS?
Answer: Yes. Information will be placed on WWW and be available from user support.
|