Connect Digest : 2012-01-09
Hide databases from users who shouldn't be able to see them
This is a long-standing request from Erland Sommarskog which I've highlighted in previous digests. But the underlying problem keeps coming up in multiple venues, so I thought it would be good to call attention to the item one more time. Some will argue that the contained database feature provides a solution for this, but that only works well if you want to restrict a user to exactly one database, and only works well if your application is compatible with the limitations of the feature. Please comment on the item and explain how this feature will help you in your environment.
Contained Database users are people too
In playing with the contained database feature as a solution to Erland's concern above, I discovered an unfortunate bug: a database-level user (with password) who has connected to their contained database using SSMS will not enjoy most of the important IntelliSense features. I'm highlighting this Connect item not so that you can vote for it, but rather just to be sure you're aware of this limitation if you intend to utilize contained databases in the short term. As an side effect, I also discovered that there doesn't exist a straightforward way to set up a contained user that can bypass the password policy in place, unlike server-level logins (where you can say CHECK_POLICY = OFF). Personally I think they got this backwards – logins are the security entity where you want to make it harder to implement simple passwords. If you want a contained user with a simple password, you can create a server-level login, associate it with a database user, and then use sp_migrate_user_to_contained (note that I haven't tried this).
Please just go parallel, regardless of other factors
Paul White (@SQL_Kiwi) has asked for an option that is kind of the opposite of MAXDOP. I say "kind of" because he doesn't want to be able to say MINDOP x, but rather try to coerce the optimizer to use a parallel plan and then follow the same rules it normally would in determining the level of parallelism.
Expose SHOW_STATISTICS through a DMV
Greg Low has proposed adding a DMV that would mirror DBCC SHOW_STATISTICS output, making it easier to work with the results. I'm all for this, as it can be quite a hassle to mix monitoring queries with DBCC calls.
Check constraints during CHECKDB
Thanks to Ola Hallengren, they are considering adding the ability to check all constraints (and, where appropriate, mark them as trusted) as a part of the DBCC CHECKDB process (specifically, using the EXTENDED_LOGICAL_CHECKS option). There are already plenty of votes, but more votes (and, more importantly, comments about how this will help in your environment) will help.