Swati Tiwari

HDI CIFS Connection and Scale?

Discussion created by Swati Tiwari Employee on Dec 6, 2017
Latest reply on Dec 6, 2017 by Swati Tiwari
  1. What does HDI call a CIFS connection?
    • If a user had 6 mapped drives, will that mean # of users times 6 = CIFS connection? So in this case 8000 users * 6 mapped network drives = 48000 connections?
    • If a user has multiple files open, does that count towards CIFS connection? I assume no but want to confirm
  2. CIFS connection is the main bottleneck so if we use 64 GB of memory per node, what is the maximum number of CIFS connections supported for optimal performance?
    • Are there any CPU requirements as well?
    • Can this support 9600 connections or 6000 connections based on release notes?
    • Is this per node or per cluster?
    • If it is per node, what happens when one node goes down?
  3. We assume 10% concurrent users so that means 600 connections?
    • Do we mean simultaneous logging or just online together?
  4. Custom version delete script
    • How will it identify orphaned versions? Will it crawl the entire HCP and delete versions one by one for each it cannot find a match?
    • Will it reference HDI and delete all other versions from HCP?
    • How will the performance be if the number of objects are large?