Last modified 7 years ago Last modified on 06/11/13 15:12:39

DevOps Weekly Meeting | June 11, 2013

Time & Location: 10am-1:00pm in LHL164


mhanby, tanthony, jpr, pavgi, billb, bade

2013-06-11 Agenda

  • clc genomics trial
    • drmaa and qsub working from node.
    • assuming issue is internal to CLC GS and gridagent launch
  • new drawing for our hardware
    • have new 10g connectivity through brocade
    • galaxy seems to be faster
    • replacement hardware
    • move rcs-srv to rust (power)
  • interns in the lab
  • storage billing model
    • labor @ 2hrs/wk
    • contract costs
    • need to learn update recommendations from Crowbar/Dell?
    • q: is there a limit to the amount of storage that can be staged?
      • if we have TB-month allocations and someone has $5k so is eligible for a 400TB-month allocation. What happens if they want to store 400TB for 1 month? is there a limit to how much they can store?
      • There are two obvious and real limits, the time it takes to stage that much data dictates how long it will take to load. Also, and more critically, we don't have 400TB available until more storage is bought.
      • idea: a max currently allocation amount.
      • issue: the storage fabric does thin provisioning so it's not clear how much storage will really be used
      • possible solution: we don't bump allocation until we get 80% full
      • issue: how do protect a user from deleting all their backups? can backups be delete protected--depends on features of crashplan, our targeted backup solution. -- need to understand capabilities of crashplan service
      • issue: what is the exit plan? what happens when someone runs out of money to store data? what happens on amazon?
  • duracloud
  • box comments
    • alternative?
  • misc updates




  • 16.png (133.2 KB) - added by jpr@… 7 years ago. Image of ganglia showing impact of gen1 hardware taken offline
  • IMG_1956.JPG (2.2 MB) - added by jpr@… 7 years ago. updated network map for rcs with cloud and current gen2&gen3 hpc hardware