Jump to content

NOW where's the bottleneck?


This topic is 7173 days old. Please don't post here. Open a new topic instead.

Recommended Posts

Hi, all. I've been logging stats from FMS3 since mid-May. We run into occasional "system" delays wherein one user's activity brings others to a standstill -- the appearance of The Coffee Cup Icon.

I have come to the understanding that this is usually due to FMS halting data distribution while that user's workstation does its crunching. Does this sound like an accurate assessment?

Secondly, I am wondering if the stats I've gathered point to a weakness in our system that can be easily/inexpensively addressed. For example...

Net peak/avg (kb/sec)

412 / 4

Disk peak/avg (kb/sec)

461 / 6

Transactions peak/avg (/sec)

1363 / 4

I have no sense of scale for these numbers against the likes of 10/100/1000 ethernet, factory ATA vs. SCSI vs. U2W SCSI, etc. At first I thought we might need to ante for a gigabit switch (we have several G4 Macs, including the user named above), but then I read somewhere else that the disk subsystem can be slower than 1000T ethernet.

The numbers above represent a jump that occurred when we ditched several 6100/7100 class machines for Beige and BW G3s. The only place the old machines compare to the new is (theoretically) the 10T ethernet, so the fact that Network, Disk, and Transactions stats went up seems to corroborate that processing power has gone up. Also, Network and Disk were not previously peaked out, since they subsequently grew -- but I can't tell whether the latest numbers represent a limitation of the server or the clients.

Do the upgraded clients mean the server has met its match, or does the server still have plenty of Network and Disk headroom?

Given that the clients are running FMP4, would the user at the beginning of this diatribe see better results from a faster CPU?

Link to comment
Share on other sites

We had a sub-summary report on 50,000 records that would slow down the whole system, so there are limits. This is true for FMS3, 5, or 5.5

Upgraded clients help, not hinder. We run FMS3 on windows NT, with SCSI HD. When the 25 client machines were upgraded to new windows boxes, FM4 "screams" (much faster overall, and maybe faster than FM5. No coffee cups)

It is possible that one slower client is holding up the others. Usually, the slowest client has trouble "getting in" to the server, as faster clients dominate the traffic.

As always, FMS needs the fastest hard drive you can afford

Gigabit ethernet is probably overkill

gregory durniak

Link to comment
Share on other sites

When and Why the Coffee Cup Appears on Screen

"One last situation is possible but is very rare. If there is a machine on the network that is slower than the rest of the guests then it is possible for that machine to get stuck trying to access the database and result in a coffee cup. This happens because each machine on the network will poll the host when it is busy. Faster machines will get in line and even though they wait the same amount of time they may be able to react more quickly and get the hosts attention. What happens in a busy situation is that the slower guest keeps getting bumped to the end of the line."

I doubt that a gb switch is the answer -- but what hardware are you running the server on? What about your current switch -- can you force the ports to run at a set speed vs. auto-sensing? What size cache is the server set to? This is discussed in FileMaker Server Best Practices. You might have to dig around for the older version.

Oh -- and make SURE FMS is the frontmost app on OS 9!!!

Link to comment
Share on other sites

This topic is 7173 days old. Please don't post here. Open a new topic instead.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.