What's A Good Ballpark Figure For Server Load For A Server?
It's Usually Around 1.06, But I've Seen It Get As High As 6.50

It's a P4 2.6ghz, 1gb DDR RAM. Redhat 8.0

I have a p3 that has spiked to 13.00 because of ensim being overloaded from a mailing list program.

Now most people would say that the server should have crashed, but it didn't. It was a good thing that the process had finished within a minute, and by another 3 minutes time, the server was below 1.00

It would do this every 15 minutes without fail.

Was I worried.....you bet.

Did the server handle it without giving up http access and mysql processes....yes it did to my astonishment.

Did I move the offending account to a semi dedicated dual server......you bet I did.

This was what I would call a good experience in dealing with server load.

I had a bad experience with someone using an older version of YABB.

That script would leave open processes that slowed the system down noticably. The server load was around 3.00, but the worst part was the fact that other users also were degraded. The server wasn't spiking to a high level, but rather staying at this level of 3.00 causing my beeper to go off constantly.

What I found to be funny in these two cases is that even though the mailing script would reach levels of server load that totally blew my mind, the server never crashed and I did not get emails of complaints......but.....when the server maintained a server load higher than normally expected for a sustained amount of time as in the case of the YABB user, I would get serious degradation in performance, complaints, and I also had to Reboot the system remotely.

So I guess the moral of the story is, servers can handle spikes of server load and if they are built well, just keep on chugging. But sustained server loads over exceeding normal operation will degrade performance.

When you find the source of the server load, simply notify the customer and upgrade him as necessary.

Robert

:emlaugh:
Thanks For The Story.
But You Didn't Answer The Question :D

The industry standard is 1 for a single cpu system and 2 for a dual cpu system.

Robert

Thanks.
Any Ideas About My PHP Problem?

Has BobMarley said but i think anything lower than 1.5-1 on a single cpu or your server is fine!

Pipson

Pipson, Do I Know You?
XAvia?

server load can vary depending on what your doing, for example if you have something like cpanel backups running the cpu might at 75%+ idle but the amount of I/O is causing the load.

A server should never run at a constant 1+ load if you have a single cpu system, sure if can burst higher for a bit but should never stay running that high.

I got a bad server
Celeron 1.7

getting often at 1-2.0

@@

What is the 5 minute and 15 minute numbers?
Those are more indicative of the true server load along with CPU usage. Servers will peak for varying reasons and every one is unique in the amount of load it will take.

I have a AMD Athlon 1700 that I have seen peak at 32 and still served pages like it was at .3

1 to 1.5 is good ball park to stay under without effecting speed.

These numbers are nothing. I've seen server loads in excess of 200 before a Celeron 1.3GHz would stop serving pages. It was running around 8 on a regular basis except for those 90+ spikes. Interestingly, I haven't seen anything near that since some software upgrades recently. It's typically been below 2 since then.

It is not true that server loads above 1 means bad. it is just how many processes that's waiting in line for processing. Obviously a higher number means that you have to wait a bit longer. But it's not of the stage where the server will burn out or similar.

Obviously keeping it always belong 1 is good. but is not necessarily a bible to follow.

Cpanel does spikes the load a bit on it's nightly cron runs. It's ok if it is just an hour or so during that time.