We've just moved all the PASS chapters that were hosted on ~.ssc.com domains over to ~.sqlgroups.com. This is part of the final cleanup as we try to make sure we don't have anything left pointing at SSC. If you're the leader of one of these groups and are hosting elsewhere please let me know, and if you need hosting for a SQL group we'll be glad to help with that too (it's free).

apac.sqlgroups.com
austin.sqlgroups.com
boulder.sqlgroups.com
charlotte.sqlgroups.com
chicago.sqlgroups.com
chicago2.sqlgroups.com
cincinnati.sqlgroups.com
claycountyhistory.sqlgroups.com
community.sqlgroups.com
denver.sqlgroups.com
detroit.sqlgroups.com
houston.sqlgroups.com
indianapolis.sqlgroups.com
innercity.sqlgroups.com
issug.sqlgroups.com
louisvillesql.sqlgroups.com
nashville.sqlgroups.com
nyc.sqlgroups.com
nyc2.sqlgroups.com
olympia.sqlgroups.com
passsigs.sqlgroups.com
philadelphia.sqlgroups.com
rosql.sqlgroups.com
sgsql.sqlgroups.com
siliconvalley.sqlgroups.com
steelcitysql.sqlgroups.com
tampasqlbi.sqlgroups.com
triad.sqlgroups.com
usergroups.sqlservercentral.com.sqlgroups.com
utahcounty.sqlgroups.com
vancouver.sqlgroups.com
westpalm.sqlgroups.com
wisconsin.sqlgroups.com

I imagine many of you have seen or attended a Code Camp, a one day free mini conference type event that focuses on demo's and code. Brian & I have done SQL presentations at most of them in Fl and they vary from fair to great events, all depending on a combination of the venue and size/experience of the volunteers running the event. They always have good speakers and topics, I think mainly because people who are passionate about their craft enjoy talking about it to others and that just shows.

But anyway, for a while we've been wanting to do a similar event that just focused on SQL professionals and after a couple months of getting stuff sorta ready, we're finally ready to go. SQLSaturday Orlando is live and we're signing people up for the email updates. Once the schedule is complete (around Aug 20 we hope), we'll get it published and then start taking actual registrations for the event in early Sep. The reason for that is that we know there is typically a 10-15% noshow among those that registered (woke up tired Sat morning!) and we want to minimize by not having people register months in advance. Actual event will be Nov 10, 2007, at the Heathrow Campus of Seminole Community College. Call for speakers is open and we're also accepting suggestions for topics. Come take a look!

If all goes well we're hoping to do the event in both Tampa and Jacksonville next year, and who knows, maybe elsewhere too!

This turned out to be a nice event, better venue than last year. Well organized is always appreciated even at free events. Kudos to Keith & Keith at VisualGov for a first class job.

Brian, Chris, and I did our presentations on Sat and all went pretty well, we gave away a bunch of books and coffee mugs mostly to make some room in my office! Only downside was that because we were either presenting or manning the End to End Table there wasn't much time to enjoy the other sessions. Had a chance to talk with the facility AV guy to ask about their projectors (because they seemed kinda large), said they were 5000 lumen projectors that ran about $7500 each. We use a 2000 lumen projector in our classroom and once in a while I look up at the wrong moment and it about blinds me, not sure anyone needs 5000 lumens. And we still had to dim the lights to make everyone happy.

Not a happy story, but if you've got a child I suspect it will be easy to decide to buy a raffle ticket to try to help out. My friend Steve Jones posted the raffle over at SSC. Brian, Steve, and I (End to End Training) are doing our bit by providing some of the prizes, and we'll be contributing some cash as well. All proceeds go to help Katie directly, nothing to expenses.

Browsing MSDN this morning and saw the note that they now have Wiki support. If done well it could be tremendously powerful to get not just the official MS line but also the hard lessons learned that you often run across in forums and community generated articles. I haven't looked yet to see if they support adding new topics, it would definitely be limiting if we can only extend existing topics.

http://msdn2.microsoft.com/en-us/library/ms130214.aspx

NewID() is commonly used in the order by to return the result set in random fashion. No idea whether its a "true" random, but usually good enough for day to day purposes. It's a decent technique and easy enough to apply, but it can be a performance killer. In the situation I saw they wanted to return a sampling of the matching universe to the user and they accomplished this via newid. The downside is that in some cases there could be 200k records in the initial match set, so you have to wait for SQL to assign the newid's to each row and then do the sort. And the wait can be more than a second or two! They had considered just adding a uniqueidentifer column to the table just for purposes of randomizing. Trade off there is that you add 16 bytes per row, and then you need to index it to make it effective, requiring more space and overhead. In the example I saw it looked like the overhead would have made sense, removing the newid from the order by reduced query time by about 50%.

I recently had the chance to look over a friends shoulder at a query that needed some tuning work. In the previous iteration of tuning they had found that running a portion of the query inside a cursor actually yielded better performance than a one query takes all set based approach. If you've read my SSC article about cursors you'll know that I'm not anti-cursor, more a pragmatic-cursor guy, so the question is - without benefit of a couple hours with the code and data - why would the cursor solution be more efficient?

Don't have a provable answer, but my guess is that the sheer size of the set based approach was driving one or more table scans. Not that big queries can't be done without table scans of course. If you think of one of the tables as having the where clause values, by looping through those and executing the rest of the query you would potentially get better query plans in because the optimizer would see that the value was distinct enough to use an index. So essentially by running more specific queries the statistics might allow you to get a faster plan than if you run it all at once.

Note that I'm not suggesting that everyone start using cursors to boost performance, just that there have been cases where it worked and it's interesting to consider in which scenarios we might look to the technique.

I just got my author copies of the Sybex book for exam 70-444 and have to say the book looks good! I wrote the SSIS chapter and it is amazing what a couple good editors can do to make it really look professional.

I've enjoyed the simpler rules and spontaneity of publishing content on SSC for years and I had been reluctant to make the time commitment even a chapter requires. This project came to me via my friend Steve who was asked to do a couple chapters and thought it would be a good learning experience for me (and it was). The editing process was rigorous, but not unbearable, and it was easy to see how the final product would benefit. I had to learn their formatting rules which made the writing just a little harder than it would have been otherwise, but maybe the most challenging and most fun was figuring out just what to put in or leave out to support the objectives. I wrote about 90% of the chapter, then went over to take the exam to make sure I wasn't too far from what MS in mind. Passed the exam decently, and as expected the SSIS questions were very high level. I haven't looked at the other 70-444 books to tell you if this one is better or worse than the others, but if you want to take a look anyway it's available at Amazon.