OK, so there are WAY too many things at the Summit to attend. It's an embarrassment of riches. A few sessions may be duds, but it's just as likely that three or more can't-miss sessions are running at the same time. So here's just a snippet about my day, where I couldn't see even half what I wanted to!
Today I hit the keynote, obviously. I'm not usually all that excited about the marketing spin and announcements. This year, though, something really exciting: column store/VertiPaq in relational databases. Wow. Game changer. Amazing. Other, better bloggers I'm sure will cover that new feature in depth, so I won't beat a dead horse here. I sort of guessed that this was coming, but it's here sooner than I expected, and I can't wait to implement it.
I then hit a SQL CAT session, set up as a panel of DBAs from larger customers, presenting HA/DR architectures they have implemented - a really great session. I love the SQL CAT content because it seems to be much more real world, no marketing, straight-story experience and less spin. Presented were several different, actual, working topologies for combining failover clustering and mirroring across data centers, with examples of minimum-downtime data migrations and upgrades. Good stuff. Directly applies to what I do.
Next was a session with Eric Hanson on warehousing best practices - a perfect fit for me, as the physical design for my organization's warehouse falls under my responsibilities. Turns out our design is pretty close to optimal, with some exceptions, which was great to see. More very exciting possibilities with VertiPaq also demo'd here.
I went to a session next with Jen McCown, whose work I know from her prolific blogging and other online content. This one was a little bumpy. Jen is funny, and good under the pressure of presenting, and got through it with grace (plus a lot of great audience participation). The biggest issue, for me anyway, was that she first went into an explanation about the logical query process, but without articulating really clearly that SQL Server's actual, physical implementation of query processing doesn't really follow the steps -- nor does it need to -- in the theoretical/logical processing model. In other words, the logical processing model is really only a mental model that we use to help imagine how results derive from a SQL statement, but the internals of the engine processing our queries and the optimizer are only bound to give us the same results as the logical model, not to literally use the same logic to produce those results. If the logical model were literally implemented, it would be far too slow to be of any use. I'm sure she knows all this, but it seemed like perhaps she had a hard time articulating it. Other than that there was great content to be had, sort of aimed at the intermediate SQL coder, and it got better as the session went on.
Last I had lined up Adam Machanic's session about auditing, and was really looking forward to it, but then, as they will, a minor production issue at work interrupted my plans and I spent that time killing a runaway insert and shrinking a log file at the office. Thank goodness for laptops and WIFI. I will hit him up later for a 5 minute version and bibliography :-).
I have two experienced Oracle DBAs attending with me this year, and I love the Summit so I've been a little paranoid about trying to be sure their experience is good. They reported the first session they picked was too simple, but then got in a groove with about 300 level sessions that seem to be a fit. Hopefully we can locate the best stuff to help them repurpose the wealth of knowledge they have already. We also have a BI architect attending, but I somehow didn’t cross paths with her today. Have to make a point to find her tomorrow.