Subject Re: Firebird Usage Load Problem
Author Maurice Ling
>
> I don't blame him for getting upset about your resource usage! The
way you
> have the cache configured, EVERY CONNECTION is getting statically
> allocated > 255Mb of RAM for its cache. A single connection really
has no
> way to use all that cache. Heck, it would even be outrageous to
allocate
> *that much* cache to a superserver, which allocates cache
dynamically and
> shares the cache amongst all of its connections. Revert it to the
default
> 75 pages, i.e. 300Kb per connection, which should be ample. Just be
sure
> that you have enough RAM on the system to support the number of
concurrent
> connections you expect.
>
> What is yomping up CPU? Disk i/o gone wild as your seriously
RAM-starved
> system juggles what's left back and forth to/from swap? Operations
queued
> up waiting for resources? Multiple processes all running similar
> CPU-intensive ops on the server? ????? First thing to eliminate
will be
> that RAM greed...
>
> ./heLen


Hi Pavel and Helen,

I've set the buffer back to 75, stopped and restart my script.
A bath... a snack... and 30 minutes later, this is the "top" output...

30342 mling 15 0 48616 44m 2980 S 0.3 1.1 0:38.35 python
30343 firebird 20 0 15000 12m 10m S 65.3 0.3 2:08.87
fb_inet_server



FB takes about too long and too much CPU to run, compared with the
calling script.

I do not know how to explain that the average load of my script is
about 0.3-1.8% CPU while FB database access is 50-85% CPU load.

Alright, when I used buffer size of 75 and running 3 scripts, the 3
fb_inet_server processes run at 3 x 99.6% CPU load while my scripts
run at 3 x 1.5% CPU load. When I increase the buffer to 40000, the FB
processes uses about 3 x 70% CPU load with no change to my scripts'
CPU usage.

In comparison, mysqld (mysql daemon) runs at an average of 1.5% CPU load.

To my system administrator, I am strongly advised to use a REAL
DATABASE SERVER. Helen, I've been corresponding with you and you know
I do like Firebird. I've even published a conference paper on it. But
then, numbers for numbers, I will need to bring FB's CPU usage to less
than 10%.

Any insights?

Cheers
Maurice