Comments (4)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 Maduskar commented Permalink

Hello We have a XIV having "Write Miss Large" with value touching 55000. This XIV is back end for IBM Dedupe appliance. Does this large value signifies any thing? We have resource contentation problem and backups get hung on Dedupe. <div>&nbsp;</div> Regards, <div>&nbsp;</div> sandeep.maduskar@equifax.com

2 anthonyv commented Permalink

Hi Sandeep. <div>&nbsp;</div> Your getting Write Misses that are larger than 64KB in size. <br /> You dont say if they are IOPS or MBps, but I am guessing its 55MBps (rather than 55000 IOPS). <br /> There is no way you can control the number of write hits (or misses) that your application creates (regardless of whether its IBM ProtectTier or something else). <div>&nbsp;</div> As for what for it means... it just means writes are being spread across a wide spread of logical block addresses. I doubt it is causing contentions or hung backups.

3 V7CW_Andy_Egemo commented Permalink

Working with the new XIV Gen3 without SSD. When looking at the Hit+Miss latency for all size data, shouldn't the read and the write Hit+Miss latency add up to the same as the read+write Hit+Miss Latency? I show the Read+Write significantly lower in the GUI stats than the culmination of the two together. Maybe I'm missing something here, but I would think that they would add up to equal the same stats as the Read+Write. Maybe a bug in the code? We're currently at 11.1.0.b microcode..... <div>&nbsp;</div> -AJ

4 avandewerdt commented Permalink

Interesting question. I am thinking about the maths here and wonder if the total percentage looks skewed because there are more reads than writes. I would need to do a careful examination of a real machine to be sure whats really going on. Might be worth getting your IBM FTSS or TA involved.