Picked this up from the local Net group here in Orlando. INETA is actually pretty helpful when it comes to SQL groups. This is definitely a specialized role and not one I'm sure I agree with, but it's hard to argue against their success at least with regards to their speaker bureau. Wouldn't hurt if a SQL person got this role!
"The INETA NorAm Board of Directors is expanding to include one additional member for a total of 10 people. The new Board position will be focused on sponsorship: - Develop / maintain a sponsorship plan for raising funds. - Determine sponsorship needs of the organization. - Work with the Treasurer during budget planning and monitoring. - Develop plans to recruit new sponsors. - Recruit new sponsors. - Maintain relationships with existing sponsors. Commitment is 15-20 hours/month with regular bi-weekly Board conference calls and two expense-paid in-person weekend meetings per year. Any INETA user group leader or existing Board member may submit a nomination. Nominations should include the person's name, email, phone number, description of why they would be a good candidate, and any supporting materials. Nominations should be sent to noram.secretary@ineta.org no later than 12:00 EDT Monday 11/5/2007. Voting by the existing Board members will be complete by Mon 11/12/2007, and all candidates notified shortly thereafter. Please refer any questions to noram.secretary@ineta.org."
After a few meetings of oPASS this year the numbers I use are 2 slices of pizza and 2 drinks per person, plus about 10%. I order the large pizzas and get a mix of crusts and topping - remember to get vegetarian! In our case we also have one member who is allergic to onions so we make sure we have at least one pizza that fits in that category. For soda it's usually something with caffeine, something without, and bottled water. It's definitely good to have extra drinks available, they can always be used next time.
Our meetings start at 6 pm and we like to try to serve the food at 6:30 so our speaker can start at 6:45 or so. That means we have to place the pizza order before we know how many will be attending. We always ask members to RSVP but in practice I take that number and multiply by 2 and that seems to work out to be close. Ideally we'd move the food to a little later in the evening so that we could order right at 6 pm and we may do that - or may not!
For planning purposes I figure its $5/person for food/soda/cups/ice/napkins/etc. I typically ask sponsors for a little more than that so we can bank a little for the less frequent but still recurring expenses all groups have.
This went up today and there is one great comment about possibly using snapshot isolation as an alternative to the techiques I described. It's the best part of SSC; reading the article plus the comments really gives you an interesting view of the subject at hand.
The article is at http://www.sqlservercentral.com/articles/Database+Design/61335/.
I've been following Steve's exploits trying to learn the podcasting business and we usually talk about it once a week or so, just me as a sounding board/podcast skeptic. We were talking yesterday about the podcasts and I had mentioned something about making a note for my blog, and he responded that he had just looked at it for the first time. I'm famous!
Well, perhaps not quite yet. I think it does point a weakness on the SSC side that they don't bring blogs into the fold as much as maybe they should. To be fair that weakness goes back pre Red Gate. Still, I'd like to see all the SSC blogs at least considered for inclusion on the front page and daily newsletter. Don't know if automaticaly inclusion would work because not every blog entry - perhaps especially this one - aren't likely to be of interest to the general community.
If you subscribe to the Connector (the main communication email from PASS.org) you should have gotten one yesterday that finally linked to the event! Looks like we finally sorted out the lines of communication and every bit of press helps.
If you do any coding in .Net using aspx pages you'll recognize this one. Each page has a title attribute that defaults to 'untitled page'. Nothing wrong with a default, but I sure wish they wouldn't let those pages compile or would at least flag a warning. All too often I run across simple sites where the title in the browser is 'untitled page'! Heck, I probably have one on one of our sites somewhere, but I try to always check. Now when you see it you'll know why it's there.
Had a friend call me recently to ask about how to do this, had an nvarchar(2000) column that needed to be unique values. If you've ever tried this you'll know that SQL will let you do - with a warning! All works fine if your column initially has less than 900 bytes for any row but the first time you try to update to 901 bytes - bam! Now I know many of you will be shaking your head about the idea of an index that big anyway and I don't disagree. At the same time it was just a business problem to solve and he was hoping that there would be a quick way to do without a lot of changes. In other words, having the column be unique was important but it was far less important how it ended up that way.
No good solutions to this. My first thought was to hash the value and put a unique index on the has. Doable, adds a bit of overhead, index would work fine. Another idea would be to create a computed column on the left 900 bytes and index it, that would work unless we really had values over 900 bytes (which is really a key question, many columns are defined far larger than the average value just in case) and wouldn't break the index even then, it would just cause us to treat some rows as unique that really were not (because we'd lose the trailing 1100 bytes). A variation of the first idea would be to move the large column to a lookup table, create a key (hash, guid, or identity) and store that in the main table with a unique index applied). But the easiest - if clunky - is to just do an exists check in both insert and update triggers.
Moving the logic to the trigger was the easist solution because it was essentially the same logic they were already using in a stored proc. They were hoping for a miracle of technology with some new feature in 2005, but even included columns won't solve this one.
We're at 215 registered with 3 weeks to the big day! This week I'll be having the last planning meeting with our facility host and finalizing the food plans (anyone know how much coffee we need for 200+ attendees?). PASS will be giving us a nice mention in their next Connector and we've asked all our speakers to reblog/recontact their lists to finish getting the word out. We're drawing people from all over Florida and a few from outside the state too!
At this point the biggest worry I have is logistics. Signs to get people to the event, signs to get them to the rooms, having enough and the right kind of food, tables for sponsors, tables for food, cups, ice, water, trash bags, check in and registering walk ins, etc, etc. We'll have about 10 volunteers at the event (each working about 2 hours so they can still enjoy the sessions) and that will really help, but I suspect the biggest pain will come from the things we just didn't think of (or thought of and forgot) in advance.
The most recent ACM Queue Magazine (Vol 5 #6, Sep/Oct 2007) had two really interesting articles. The first was about the ZFS file system and was framed as an interview it with the two inventors, both from Sun. I'd seen it mentioned a few times via blog and took the opportunity to learn more. It was interesting to hear that they built the file system to expect hardware errors, to never require a disk check, that they think storage really should be a pool that you can just add (much like a SAN, but at the file system level). They talked about the fact that all drives have defects at start and that errors are always occurring, its just a matter of which ones get passed up to the OS. You can see the article at http://www.acmqueue.org/modules.php?name=Content&pa=showpage&pid=504&page=1. The second article (same issue) was about HDD's and was written by Jon Elerath of Network Appliance and really went into detail about how drives map and remap to handle errors, how the SMART functioin of drives isnt necessarily that smart, and some other good stuff too! More and more DBA's are insulated from storage but I think its a good idea to have a least some of idea of whats happening in the storage industry and how it relates back to us.
I wrote this article shortly after returning from the PASS 2007 Summit and tried hard to come up with ideas to address what I see as shortcomings within PASS. PASS runs a great conference, but in other areas they just haven't achieved the kind of success that I know is possible - I look at SSC as an example of what is possible to accomplish with a real community! I'm not suggesting that PASS should attempt to become SSC; I think community sites like SSC, SQLTeam and MSSQLTips.com make a lot of sense and fill specific niches within the business. Rather, I look at SSC with as of today more than 500k registered users and think 'PASS could have 500k members too!'.
I hope you'll read the article and take time to add a comment. Agree, disagree, or add a new suggestion - give PASS some feedback on the direction it should take. You can also see an editorial Steve wrote about the article in the Oct 18th newsletter.
All of these are new articles that have gone up on SSC recently. The troubleshooting one was my attempt to get people to think about troubleshooting before it's needed. It's much like debugging; those that are good at it typically get to the resolution much faster than those that tackle the problem without benefit of a plan and experience. The ones on replication delivery options are less interesting but good tactical knowledge.
Bill Graziano (main man at SQLTeam.com, consultant, and VP of PASS) was in town recently to speak at both the Orlando and Jacksonville user groups as a guest of our company, End to End Training. If you attend a user group you know that gettings speakers is tough, and that for whatever reason local speakers are not as interesting as speakers that visit from other places (Bill is from Kansas City). It was a marketing expense for us of course, we got a 30 second sound bite during his presentation, but it was also some good karma on a number of levels and in Orlando actually drove our attendance to a new record, 28 people!
His presentation was "What I Wish Developers Knew" and if you're a DBA you can easily understand how you might build such a presentation after a few years of fixing 'mistakes' by developers. Bill made it an entertaining evening by not just talking, but by showing scenario after scenario with things like error handling and transactions and challenging the attendees to figure out what would happen in each case. As always there were some really good questions from the audience and Bill did a great job at handling those too.
We wrapped up the session with a general discussion of SQL topics and got one really really great question - why don't DBA's and developers get along? Bill & I had slightly different takes on it, and I plan to explore it in more detail in an article soon because it really is a discussion the entire industry should have. We also got to talk about the recent PASS Summit and answer some questions about how it compared to VSLive, Teched, etc.
For those wondering Bill didn't charge us for the trip down, we just covered his expenses - so it was 3 days of lost revenue for him while he did his thing supporting PASS and the local chapters. If you can find the budget for travel, call Bill, he's a great guy, great speaker, and the members will enjoy his presentation.
I'm catching up on my reading today and ran across an article from Kathi Kellenberger called Understanding the Difference between Owners and Schemas in SQL Server. Good short overview of the concepts, but one comment really struck me when she said that she thinks of schemas as containers. That's a lot different than the kinda bland concept of owners in SQL 2000 and had me mumbling to myself again that I don't understand why MS doesn't show schemas as folders under tables, procs, etc, to make it really easy to see (and move if needed) what items are in which schema.
For all that, at this point Im still a one schema fits all person. Maybe if you have 10k tables some logical division makes sense, I've just never worked anywhere that it seemed like it would solve any particular problem. Of course it worries me slightly that Im not getting some crucial point - can anyone enlighten me?!
