Reputation
5,312
Top tag
Next privilege 10,000 Rep.
Access moderator tools
Badges
23 62
Impact
~272k people reached

1h
awarded  Popular Question
2d
comment Why wouldn't statistics update if all rows in a table were replaced?
@Mark how are you calculating the modifications?
Aug
28
reviewed Close Normalize to 3NF Question - Postal Codes/City
Aug
28
reviewed Approve User is An employee. Role or relation
Aug
28
reviewed Approve Entity relationship problem
Aug
28
comment Tell SQL Server a query is ad hoc and not to optimize based on it
i'd go with "... shouldn't be so concerned" because you still need to exercise a good deal of judgment when sifting through suggested/missing indexes. that missing index suggestion doesn't take your whole workload into consideration - even if it is part of your regular workload.
Aug
25
comment SQL Server 2008 R2 - Two Files Stuck In Restoring Status
sounds like you may have either momentarily or permanently lost access to the location of your backup file.
Aug
19
comment Statistics disappear after incremental update
per the comment thread under my answer, are you absolutely sure you need incremental stats?
Aug
19
comment Statistics disappear after incremental update
Yes. It's certainly a problem, or rather, a solution in search of a problem. I guess the solution that incremental stats provides is that you can a) reduce stats maintenance time windows or b) get higher overall sample rates because you can sample at a higher rate per partition and then resample at that rate in ongoing maintenance. However, that requires deft use of sys.dm_db_stats_properties_internal() [UNDOCUMENTED-USE AT OWN RISK]. Regardless, we don't use incremental stats on our VLDBs yet because it's not clear that the maintenance effort would result in a performance payoff.
Aug
19
comment Statistics disappear after incremental update
Thanks! It very well could be a bug, but the more I think about it it'd be quite complex to handle. Rebuild the index partition at the 100% sample rate, grab the sample rate of the stats object and resample that partition at the same rate. Ensuring that it's the same sample rate could be difficult. Alternatively, they could just rebuild the stats partition pages with the same internal logic that they use with RESAMPLE ON PARTITIONS(n) I guess.
Aug
19
answered Statistics disappear after incremental update
Aug
19
reviewed Edit Statistics disappear after incremental update
Aug
19
revised Statistics disappear after incremental update
improved formatting, corrected spelling. Thanks!
Aug
18
reviewed Close Connecting to SQL Server using PHP, ODBC, and Windows authentication
Aug
18
reviewed Reject database-link tag wiki excerpt
Aug
18
answered Is it poor practice to aggregate data from different tables into one?
Aug
17
comment Why does SQL Server Management Studio crash?
did you look at your windows event log to check for clues?
Aug
13
reviewed Leave Open ORA-19809 & ORA-12560
Aug
13
comment Should the index on an identity column be nonclustered?
It's going to depend on things like: a) workload (OLTP? OLAP? etc?), b) table size(s), c) normal form, just to name a few. You've not provided details regarding any of these factors so any recommendation would be based on guesses from your environment. Also, have you tried profiling the queries that you're proposing (with cleared buffers) and getting the specific IO profiles per configuration and seeing for yourself?
Aug
13
comment Should the index on an identity column be nonclustered?
I'm not sure that your response to @ypercube satisfies any of the criteria cited by Mr. Kejser - at least with the details you've shared. In its current form, I'm not actually sure that this is going to generate a useful answer because it would have to cover nearly every single scenario - which is already done in the blog post that you cited. If you can provide more details about your specific scenario then maybe some of the knowledge in the post can be applied.