K
Keith Sudbury
Guest
Sorry for the short hand... checking in between normal work 
Your -omsize parameter needs to be increased to reduce the OM latches knowledgebase.progress.com/.../20398 You could benefit from setting lruskips and looking at the size of your -B knowledgebase.progress.com/.../000031150 Setting lruskips to 20 will potentially give you some relief without much risk of degradation. Using a properly sized secondary buffer pool (-B2) and assigning the proper objects can also help out drastically. It is possible that the BI file isn't sized properly and/or your -bibufs parameter is too low. Most people suggest using a biblocksize of 16KB and a large enough cluster size so that your checkpoints aren't too frequent (at least a few minutes apart). Some of the MTX latches are just part of deleting index entries though. You say you have two storage areas... for the entire database or just this table? Hopefully they are at least Type II areas, otherwise you need to rethink the physical layout.
Continue reading...
Your -omsize parameter needs to be increased to reduce the OM latches knowledgebase.progress.com/.../20398 You could benefit from setting lruskips and looking at the size of your -B knowledgebase.progress.com/.../000031150 Setting lruskips to 20 will potentially give you some relief without much risk of degradation. Using a properly sized secondary buffer pool (-B2) and assigning the proper objects can also help out drastically. It is possible that the BI file isn't sized properly and/or your -bibufs parameter is too low. Most people suggest using a biblocksize of 16KB and a large enough cluster size so that your checkpoints aren't too frequent (at least a few minutes apart). Some of the MTX latches are just part of deleting index entries though. You say you have two storage areas... for the entire database or just this table? Hopefully they are at least Type II areas, otherwise you need to rethink the physical layout.Continue reading...