Five Things SQL Server Needs To Drop

calendar May 13, 2010 · SQL Server Thoughts  ·
Share on: linkedin copy

While I wasn't explicitly tagged for participation in this meme, Wendy Pastrick (blog | twitter) was kind enough to tag everyone else who hasn't been tagged already.  Being in need of a first topic for a blog post, this seemed like a good one to start with!

Since a lot of things have already been spoken for, here's 3 things I particularly hate accompanied by 2 favorites from others.  If someone else has already mentioned my 3, my apologies for apparently not reading your blog!

  1.  View Dependencies
    Once I discovered this "feature" I quickly learned that it wasn't reliable, so why include it in the first place?  I say either fix it or get rid of it altogether.  I always got along just fine by querying system tables or system views.  Red Gate SQL Search is also particularly helpful.

  2. XML Indexes
    I have nothing against XML, and I fully understand that there will always need to import/export XML to/from a database.  What I really hate though is the idea of storing data in the database as XML.  The ability to create indexes on XML data types only encourages this behavior, so it really wouldn't break my heart to see it go.

  3. Indexed Views
    I understand the reasons for having indexed views and believe they serve a purpose, but I've always thought that the list of restrictions on them was rather comical.  If you're going to take that much functionality away, why bother including it at all?

  4. RECOVERY being the default RESTORE option
    This comes from Paul Randal (blog | twitter) and I couldn't agree more.  Even though I'm well aware you need to specify WITH NORECOVERY, I've still managed to bork a restore every now and then and have seen others do the same.   Having to do an extra RESTORE WITH RECOVERY at the end of a restore sequence wouldn't bother me one bit.

  5. Database Diagrams
    Courtesy of Brent Ozar (blog | twitter).  Not that I haven't used these, but I agree with Brent's statement that it's an optional feature and really don't see the point in putting smack at the top of the object explorer tree.

And that's it!  I don't believe there's anyone left to tag at this point, but if you haven't already written a post about this and want to, go for it!