FMS8 network speed

[ccosner]

Is anyone serving from FMS8 or 9 on a 10bT network? (I'm not, but a client might.) If successfully, could you describe the specs of the server and clients (speed, essentially)? Thanks.

[Steven H. Blackwell]

You'll get much better results with 100 BT.

Steven

[Vaughan]

An order of magnitude better in fact. :B

[ccosner]

Yes, of course. But has anyone actually tried it?

[Ender]

It will work, it will just be noticeably slower in network intensive tasks, like sorting and loading graphics or complex layouts.

[Tori Mitchell]

Ender is quite correct. FMS 8 works in a bit of a round robin fashion with many tasks. This said, anything other than a large solution should work okay... not great performance wise... but okay.

I've got a fairly complex solution that I develop for a client which has tons of relational stuff and loads and loads of graphics. I threw the switch into 10bT mode, (thank goodness for smart switches), and found that I *could* use the solution. I just really didn't want to in that state.

You'll want to limit the number of joins that you traverse in a 10bT environment as these make finds and sorts much more netowork intensive. Couple this strategy with the pre-load of images and whatnot and you should be good to go.

Keep in mind though, 10bT is ~1.3 MB/s. That's really not a lot of room on a rich client/server environment.

All that aside, give it a whirl with a pre-built solution. I think that you can download SeedCode from their website and test it. That's about the average complexity of many of the solutions that I see out in the field. If it works well enough for your purposes, by all means, go with it.

[Søren Dyhr]

You'll want to limit the number of joins that you traverse in a 10bT environment as these make finds and sorts much more netowork intensive. Couple this strategy with the pre-load of images and whatnot and you should be good to go

Could you expand here, "joins" is too loose a term for me the get the fuller picture, do you here mean the use of the separation model or similar, with different relatioal structures in the two ends - or??

Are you ahead of the rest of us - in utilization of this:

http://www.cnsplug-ins.com/products.htm?product=mmquery

Please enlighten us here!

--sd

[Vaughan]

Put it this way: I set up a FMS 8.0 hosted solution that has users in two cities some 600 km apart in Australia connecting across a (T1?) WAN, and speed is "satisfactory". Finds are slower for WAN users than LAN users, but only on unstored fields, and only the first time until the data is cached. A process that takes 4 seconds to complete for local users also takes 4 seconds for remote users, so by and large it's awesome.

Let's also remember that 10baseT was the original Ethernet and it was the standard until just a few years ago.

[Tori Mitchell]

Joins = a relational line in the TG. It's the way that you referr to a relationship like we use in FM out in the SQL circles.

Sorry, I was only referring to the way that filemaker deals with relational traversal across TO's. If you're traversing more than three or four, (and four is a stretch depending on the way that things are joined), you're in trouble in FM in *general* due to the way that the network activity in FMS 8 works.

hence:

Company --> Regoin--> District --> Branch --> Job --> Employee --> etc...

Is overkill for many instances in FM, at least in regards to what you should be working with all in one shot, ne?

If there's a igger question to what I'm talking about, we can handle that in a PT.

FMS 9 OTOH, well, that's a different post altogether isn't it?

[Søren Dyhr]

Don't forget manymany recursive structures then, it's where it really can be nesting deep!

What would you suggest to someone who thought that chapter 10 of ISBN 1-55622-859-7 was the answer to thier prayers of a way deal with inventory levels, when sets of items are very likely to occure in a beautiful mixture single items.

--sd

[Tori Mitchell]

I'm not really sure that I understand the question but, then again, I've never read the book in question.

The problem of inventory levels is an interesting question though, I've found that the only thing that can be done about that is to provide a good toolset for entry and subtraction via system usage and manual updates via cycle-counts and manual, whole-hog, yearly inventory.

What specific details of the inventory process was the book addressing that would deal with more than a simple log of additions/subtractions?

[Søren Dyhr]

What specific details of the inventory process was the book addressing that would deal with more than a simple log of additions/subtractions?

Pretty much the same which is covered in the second half of this:

http://jonathanstark.com/recursive_data_structures.php

--sd