Comments (3)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 localhost commented Permalink

I have been attempting since 1994 to come up witha comprehensive methodology to do this.I think the three main areas are:1. Sub-Channels2. CP Utilisation3. Memory UtilisationI have never had problems with 1 or 2, but I expect them to eventually become a bottleneck.Memory is interesting, and comes in three flavours:1. Usage within structures2. Usage by structures3. Back-Up & Dump -- some of this is mitigated by systems managed duplexing. And, dumps can just take over the 'free space'.

2 localhost commented Permalink

<p>Hi Ted! Thanks for your comments. I've heard the term "white space" used a lot for what you called "Back-Up" if I understand you right.</p> <p>I think we understand there probably are some shortfalls in instrumentation. But I'm hoping we can do a lot with what we have.</p>

3 localhost commented Permalink

Agree with Ted. Recently having some issues here and Mike Fast analayzed some of our CF / XCF reports. Some ROTs on path busy, subchannel delay, CPU utilization etc. would be nice. The XCF performance flash / white paper is nice so perhaps some of that could be incorporated. Mark Brooks also did a good session on XCF signalling at SHARE in NY last summer that would be good information to publish more formally.