SQL 7.0 SP1 on NT 4 SP6a Cluster.
We have a DB with the option "select into/bulk copy" set to true.
When non-logged operations happens, the backup of the transaction log fail, so we do a differential backup instead and everything is ok.
My questions are:
UPGRADING to SQL 2K,
1 Do we have to choose the "Bulk-Logged" recovery model?
and if so,
2 Do we still need to run differential backup because backup of t-log will fail as on SQL 7.0?
Can somebody help me with this?
TIA.
Franco
:cool:In BULK_LOGGED recovery mode certain bulk operations are only minimally logged increasing performance and decreasing log size. Performing builk operations do not require you to process backup, since they are actually logged thoughout BCM page. For every datafile there is one BCM page where one bit correspond to extent modification by minimally logged operation. When you perform tran log backup, sql scans BCM and backups these extents. As a result you are getting larger tranlog backups then in FULL mode.
HTH,
OBRP
Showing posts with label operations. Show all posts
Showing posts with label operations. Show all posts
Wednesday, March 21, 2012
Friday, March 9, 2012
non clustered index on heap
What would be the implications in terms performance and i/o operations, if a non clustered index created on heap ie. without a clustered index on table.It depends. There is no straight-forward answer. It depends on the schema, row size etc. Generally speaking, it is recommended to have a clustered index on every table especially so for large tables. There are however cases where you can get the best bulk insert performance by inserting into a heap vs clustered index. Also the more indexes you have on a table slower the performance. Search in MSDN for the whitepaper of bulk load that should give some ideas on one aspect of this problem.
Labels:
clustered,
created,
database,
heap,
implications,
index,
microsoft,
mysql,
operations,
oracle,
performance,
server,
sql,
terms
Subscribe to:
Posts (Atom)