>>I have a network that goes down from time to time, usually when someone overfills a drive. My users have been instructed to close out their apps before lunch and leaving for the day. Sometimes we still get zapped. The cure seems to be packing the large databases, however if I pack across the network, the process is very very slow. I have a job that runs at night to back up files to my hard drive, in a rolling fashion (3 days worth).
>>
>>
>>2 questions. Is it dangerous to pack across the network? (everyone is instructed to be out of the app, no guarantee though). Must you have all your tables available when you pack, or does packing just confine itself to the current table? Considering pulling backup into pack on a nightly basis, and overwriting back to server.
>>
>>Times:
>>Transact table (I DONT like this name, BTW) 10 indexes
>

Network               Work Machine (160 meg ram PII400)  Home Machine 256m 700
>2 hours 45 mins.      1 hour or so                       8 minutes

>>
>>Tom
>
>Tom,
>
>Is the server a fast machine? Can you write a program that lives on the server itself and packs "locally" on the server?
>
>How often you pack depends on how much space and how many deleted records are added each day. For large tables, many developers BLANK out the data in the deleted records and then recycle them instad of adding new ones.
>
>But, adding some space to the server may be the simplest of all, and lest costly than recovering from crashes!



Much depends on the factors mentioned by Cindy. One of the primary apps I work on has the client doing this every Friday. We also went the autoserver route so that indexing, packing and now many compute intensive operations all have their data work done on the server.

Steve