| Title: | AdvFS Support/Info/Questions Notefile |
| Notice: | note 187 is Freq Asked Questions;note 7 is support policy |
| Moderator: | DECWET::DADDAMIO |
| Created: | Wed Jun 02 1993 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1077 |
| Total number of notes: | 4417 |
I am working with a partner who is seeing poor performance
when their system is configured using AdVFS as opposed to
UFS on 4.0b. They measure throughput based on the number
of mail messages that can be written by one process and
read by another in one second. With UFS the number is 35.
With AdVFS the number is 17. The threashold value has been
set to 32K and the read/write blocks to 256.
AdVFS is a requirement here since it provides Journaling.
It's recovery mechinism is the real requirement. LSM's
not going to provide the same thing, right?
I understand that AdVFS requires more overhead than UFS
but 50% is not the correct percentage. Does anyone know
the percentage?
The application is writing thousands of small files on a
4100 running 4 processors to a HSZ40 with three striped
disks each hanging off their own bus.
Are there any other tuning parameters that could make
a significant difference in this configuration?
Thanks.
Pat Bryant
Software Partner Engineering
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1055.1 | KITCHE::schott | Eric R. Schott USG Product Management | Wed Apr 30 1997 14:49 | 16 | |
Hi There are many other things to check... the size of the advfs cache? Did you sync the files (on UFS), or did they only cache?) a single disk domain or multi... In short, we don't know quite what your test is actually measuring, nor do we know much of the system configuration. My first guess is UFS never flushed the buffers, where advfs did its log writes | |||||