http://www.sqlservercentral.com/blogs/steve_jones/2005/12/07/denver-launch/ Printed 2014/10/21 08:21PM
Denver LaunchI wrote a little in the editorial, but I wanted to drop some other observations here.
First, apologies for not blogging more, but I've been buried with kid stuff. Had my son move to a new school and it added much confusion and uproar in the home schedule.
At the launch, it was a beautiful 20F Denver day with snow falling as I drove downtown. At lunch I sat in the window and enjoyed the view from inside a nice warm building behind 20 ft windows looking out over the mountains, which I couldn't see, but it was gorgeous. If only it were Christmas...
There were at least 600-700 people, maybe more in the keynote, which was mostly a very contrived demo. They did have one local guy talk about his multi-terabyte warehouse, but his delivery was horrible and I ended up leaving. It did seem to me that most of the time was spent on VS 2005 and BizTalk 2006, but I did miss the first half hour.
After that I wandered around the expo, which was very small compared to PASS and about the size of a single major booth at Tech Ed. Still it was good to see some local companies, especially consulting companies since it's always good to make some contacts at things like this. Since I saw Mike White from Boulder and PASS and then Bill Wunder, I missed the first breakout. I did get to the second one on why to upgrade to 64bit and I have to say that I don't see any reason unless you are really pushing the limits of your 32 bit server. the presentation was lots of facts like "we've converted customers and they're 10x faster, but no detail or data to back it up. I swear these presentations are getting less and less worth listening to because they are so full of marketing and afraid to admit anything. Without the prizes I'm not sure most people would have stayed.
Afterways I talked with the speaker and he was more knowledgeable than I expected. I asked him specifically about performance in testing with similar RAM and he said it's still faster with the 64 bit processors. He attributes the better performance, especially in IO bound systems to the wider bus throwing data around quicker. I need to get back with Unisys and I think design some tests and go try this. If anyone has ideas, send me an email.
I asked about hyperthreading, actually a few people and they said it depends. So no good guidelines for determining this. I think I need a rant on the site to get someone to listen and actually publish something. Another HP guy chimed in and they said that the new CPUs have dual cores, so they are not hyperthreaded, but 2 cores together mainly for licensing purposes. Since SQL Server is by CPU, you can consolidate and save some money with 64 bit.
They also gave me a reason that made sense as to why the IA64 architecture is better than the AMD, but they expected AMD in the next version to implement a similar scheme. Sorry I don't remember, but I'm just not smart enough to get that.
I had to leave to grab kids, so I stopped back by the experts booth and fielded a few questions from people walking by with Bill Wunder. We had some advice for a few people though a guy with an AS question had us stumped. Fortunately Mike was nearby to bail us out.
One guy had an ? about a reporting server and using replication v mirroring. We told him that mirroring isn't a readable db, but mirroring + snapshot makes more sense than replication from an admin point of view. Another good debate and paper to get out there.
For a free event, it was a great time. Not like the conferences, but Microsoft put on a great time.