We are currently having some performance problems on an AlphaServer 2100
5/300 (single cpu, 512MB) where a lot of processes seem to be very CPU
intensive and take a very long time to run ie. 16 hours to process 4500
database records. This seems pretty outrageous (its canned software),
but am wondering what I can do initially to make sure the system is
tuned to handle CPU intensive operations most efficiently. Here is a
sample of vmstat output when one of these jobs is running:
Virtual Memory Statistics: (pagesize = 8192)
procs memory pages intr cpu
r w u act free wire fault cow zero react pin pout in sy cs us
sy id
3 68 23 21K 32K 9331 3M 567K 774K 22K 781K 1214 32 699 348
15 3 83
3 68 23 21K 32K 9331 203 4 173 0 8 0 4 1K 431
94 6 0
3 68 23 21K 32K 9331 46 0 45 0 0 0 7 1K 392
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 2 1K 293
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 3 1K 346
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 7 1K 353
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 5 1K 264
93 7 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 4 1K 396
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 5 1K 386
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 2 1K 278
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 3 1K 345
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 7 1K 291
93 7 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 4 1K 325
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 3 1K 248
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 7 1K 366
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 2 1K 355
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 5 1K 327
94 6 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 10 1K 411
93 7 0
3 68 23 21K 32K 9331 45 0 45 0 0 0 5 1K 397
94 6 0
3 68 23 21K 32K 9337 65 6 46 0 2 0 5 1K 311
94 6 0
3 68 23 21K 32K 9337 45 0 45 0 0 0 7 1K 430
93 6 0
I'm wondering if there is a FAQ specifically related to performance
tuning DEC
Unix (Tru64 now). Something that would give me some guidelines as to
how to tune for various types of system workloads and what key
indicators to look at when reading vmstat & iostat etc.
TIA,
clos_at_nhb.com
Received on Sat May 01 1999 - 13:14:22 NZST