A way to see if the logging is the issue might be to check the size of the log after each test - making sure to truncate the log prior...

By exporting the data, I meant exporting each table to flat files - delimited or fixed, compressing and copying those - not compressing and copying the database files. ...

If network bandwidth is most critical, you will most likely have to export the data to local disk, then compress, push across, uncompress at the destination, then load.

Hopefully it will...

congrats...kudos....thanks,

jg

I totally concur with hakan. Here is another thread dealing with a similar design question. It did not include the middle assign table, but I think might be...

LOL, it has been quite a day so far - 2 bad responses on my part...it has definitely taught me to read the OP carefully - and not in a...

oops, you are right, I answered a different question...sorry

jg

Lookup the ROW_NUMBER function in BOL. It should do the trick for you.

jg

Of course it depends , but I would look lean towards the separate tables. A bit hairier to manage, but most likely easier data maintenance, and possibly performance....

I would recommend adding a step to Jan's solution.

run the CHECKDB without the REPAIR_ALLOW_DATA_LOSS first - the REPAIR_ALLOW_DATA_LOSS should send shivers down your spine :w00t:

then review the output from the...

You are getting a duplicate PK error? My guess is the rowguid is NOT the PK - merge creates a unique index to enforce the rowguid.

after a merge, the...

I believe the uniqueness is based on the MAC address of the NIC - those should be globally unique. We use merge with approx 500 subscriptions and have had...

I would recommend using the SQL2005 system view sys.master_files rather than sysaltfiles - if you do not need backward compatibility with 2000.

jg

Mr. Moden,

your moralizing was better than mine...:)

jg

I went back to get the timings from our tests. I was mistaken, this was a financial_transactions design not inspections ( this was 1.5 years ago ;)). There...