UKStratfordMike
New Member
My background : Progress V8 (and before)/ Unix. Some DBA knowledge. Ancient IT guy brought out of (happy) retirement to help a company with performance problems with new app. Being told all sorts of conflicting things by suppliers.
Site : New Dell Server 16 procs (4 dedicated to Progress env) , 4Gb physical memory (6Gb virtual) , SAN, Windows and VMware
Software house's package running under V9.1D
Symptoms:
1) When Progress broker kicks in, uses around 25% CPU then beds in at around 5%
2) During day – see big spikes in CPU utilisation. Have tracked these down using Promon to ODBC SQL queries and Crystal. (Incredibly high request/DB read ratios). Can fix this
3) Overnight batch application devours CPU, low IO and runs like a dog. Eats around 45%-50% of processor availability
memory – using physical acc to vmware
Supplier input
1) Vmware and infrastructure confirm all set up OK
2) End application supplier (software house) suggesting it's SAN, VMware, but not their App. They set up Progress parms
Diagnostics checked so far
APWs (X4) running 99%-100% - AOK. BIW similar. No checkpointing.
Where now?
Seem to remember things like un-indexed temp tables causing massive hike in CPU util.
Has been a suggestion that Application is riddled with triiggers but we have no visibilty of source and I'm not sure it it's relevevant
Can you help?
Any ideas what's burning cpu and how can track?
Site : New Dell Server 16 procs (4 dedicated to Progress env) , 4Gb physical memory (6Gb virtual) , SAN, Windows and VMware
Software house's package running under V9.1D
Symptoms:
1) When Progress broker kicks in, uses around 25% CPU then beds in at around 5%
2) During day – see big spikes in CPU utilisation. Have tracked these down using Promon to ODBC SQL queries and Crystal. (Incredibly high request/DB read ratios). Can fix this
3) Overnight batch application devours CPU, low IO and runs like a dog. Eats around 45%-50% of processor availability
memory – using physical acc to vmware
Supplier input
1) Vmware and infrastructure confirm all set up OK
2) End application supplier (software house) suggesting it's SAN, VMware, but not their App. They set up Progress parms
Diagnostics checked so far
APWs (X4) running 99%-100% - AOK. BIW similar. No checkpointing.
Where now?
Seem to remember things like un-indexed temp tables causing massive hike in CPU util.
Has been a suggestion that Application is riddled with triiggers but we have no visibilty of source and I'm not sure it it's relevevant
Can you help?
Any ideas what's burning cpu and how can track?