With only a few days to go until people arrive in Seattle, I should probably explain what my sessions are going to be on. You know, in case you hadn’t thought to go to the PASS site and read for yourself (and for those who want to hear something that’s a little less ‘abstract’y). A few people told me last year they were disappointed I wasn’t presenting, so if you’re in that situation, maybe this post is of interest.
This year, I’m giving three different presentations – two regular Summit sessions that are on the schedule, and a 20-spot that I’m doing (twice!) at the Microsoft booth (Wednesday at 1:45pm and Thursday at 1:15pm).
The 20-minute spot is about techniques you can use to avoid data movement in MPP systems such as PDW or SQL DW. It focuses on Query Optimizer things that I do in regular SQL environments, leveraging things like join redundancy, contradiction optimisations, and tuning aggregations. It’s going to be quite fast-paced, as we have three significant things to explore, with heavy use of SQL query plans and showing how those techniques apply to MPP distributed plans. I’ve used these methods to make queries run WAY faster in both MPP and non-MPP environments, and people have said things like “I wish I’d known that before my last client” (someone at Microsoft), and “Oh, that’s cool – I’m so going to use that” (someone at a major US-based PDW partner). So yeah – come along. Both times!
The first main session I’m giving is one of my favourite sessions, and is called “A few of my favourite query plan operators”. This talk is going to focus on four Query Plan operators, but also discuss what’s going on in about six others. Or maybe seven – it depends how you count them, and how puzzled the looks on people’s faces go. We’ll explore what’s happening as your query runs, and why sometimes the counter-intuitive option could work out better. Lots of demos, as you’d expect from one of my presentations. Some live typing, and plenty of hand-waving as I describe why a particular plan shape is really what you’re after – even if the estimated cost might suggest otherwise.
The new session is one that I wasn’t expecting to give, so that’ll be fun. It’s on “The Power of Composite Indexes”, and already I’m kinda regretting it (no, not really) because Americans say that word differently to how I say it (for me the longest syllable is the “com”, while for Americans it’s the “pos”). But however you say it, the idea is to look at indexes with multiple keys, and look at how powerful that can be. And how you can kill the performance as well. We’ll be looking at issues like sargability v residuality, blocking plan operators (especially for people that came to my first session), partitions, and even helping you use T-SQL to fix one the most frustrating features about the way that queries run.
So anyway – those are the sessions that I’m giving at the PASS Summit this year. Hopefully somewhere in all that content there’s something you haven’t heard before.