How big is too big

[Ed Bradley]

I am new at large databases, but familiar with record counts below 20,000.

The FM site states that 64 quadrillion is the max number of records, but importing 4 million does not work. So I imported 41 smaller files.

Now I can not sort those without the program quitting and recovering is taking a real long time. I am beginning to think that this project is not going to work using FMP. Is it my ignorance or reality? If FMP states that a file size of 8 terabytes is possible then it must be me. Mine is 4 gigs.

I could use the 41 files individually. Would building relationships with 41 file tables be faster? Will I have to build 41 different web pages for users to access them with a browser?

I really think that I bit off too big of a project.

All help is greatly appreciated! or pointing me to a guide of some kind.

Thanks

Ed

FMSA 10 on a MacPro quad 2.66 with 4 gb ram, FMP 9 used to access files

[Fitch]

I have never tried importing 4 million records. You are pushing the practical limits of FileMaker, never mind the theoretical limits.

Do you have auto-enter options on some fields? Try disabling. Do you have indexed fields? Try deleting indexes. Calculated fields? Try commenting them out. How many fields are in the table?

You would have to design very carefully to make such a large record set usable. E.g. an unstored calc could bring everything to a grinding crawl.

41 related files, depends how they are related. Different web pages for each file? Probably not, again it depends how they are related.

[Vaughan]

What are the specs of the machine running FMP client?

Make sure FMP 9 is patched up to the latest v3. Also go into the application preferences and increase the size of the cache, that might help.

Try turning of the indexing on all the fields before doing the import, then turn them back on afterwards.

[Ed Bradley]

41 districts need to access through a web browser and pull up a list consisting of about 5-900 addresses. These 41 districts would have 100± of these lists each. They would be found by district and group listed in the coresponding fields of the address table (3.5 million addresses)

There are just a few calculations used to separate the street and number in order to sort the list by street name.

I have already imported the files and just doing a sort is minutes long and certainly will trip time outs in a browser.

Maybe break it down into 41 databases or tables. And let the district managers access independent webpages.

Or should each access the server with their own FMP license instead of using php?

Thanks

Ed

[Ed Bradley]

specs of client are 2.0 ghz, 2gb, Leopard connnected w enet cable

Also FMP 10 will arrive tomorrow if it makes a difference.

If I have the managers get FMP to remotely access server specs could vary widely.

Thanks so much for your help.

Ed

[IdealData]

access through a web browser and pull up a list consisting of about 5-900 addresses

That's a mighty big list - try a portal approach. If you filter it too it will at least by manageable by humans.

FWIW I'm with Fitch on this - maybe it really should be on SQL Server as there is so much remote access vie web browser.