Hi,
I am using BULK INSERT to copy data into a table with about 130
million rows. Typically between 100 and 5000 rows are being inserted
per bulk insert. There is usually about one such operation per minute.
Everything has worked fine for years.
Suddenly the average disk queues on the drive hosting that table moved
up to values of 50 and more, and now inserting 1000 rows takes up to
three minutes, while this took less then a second before. I did the
basic checktable and showcontig diagnosis and everything looks fine.
Definitly no schema changes.
Incidentally there was a RAID error on that machine. The table is on a
RAID 5 array and one disk had to be replaced. Rebuilding the disk took
12 hours and after that we thought that everything was fine. But:
there is another table on the same virtual disk and it is being fed
very similary without any problems. Sure enough our administrators do
not see anything wrong on the RAID array.
What I will try this weekend is to ex- / import the table data just to
rebuild it completely - maybe on a different disk. However, I would be
most thankful for any hint on the mysteriously slow inserts.
Regards
DC