Another excellent conference from Selena!
I went to almost all of the talks I intended; there wasn’t a dud in the bunch. There were others I wanted to see as well (in particular Josh William’s “Monitor the Heck Out of Your Database”) but I don’t have the ability to be in two places at once yet. Fortunately, there is video feed; unfortunately, it requires silverlight, so I’ll have to wait until the video’s ported to its final location. Meanwhile, slide decks are available for most talks on the PostgreSQL Wiki.
My top three recommendations are:
– Vanessa Hurst’s Honey I Shrunk the Database, in which she discussed copying, slicing, and mutating your data. Which sounds a lot like what I used to do with DNA (which is, after all, also data.) I also learned that you can pass multiple tables to pg_dump -t (sometimes, the best bits are those little side tips you pick up in people’s talks.)
– Stephen Frost’s Finding Slow Queries and Fixing Them. Top tip for me from this talk: if you have slow DELETEs, try indexing FKs that depend on that table – otherwise you’re doing sequential scans on the tables affected by any CASCADE options.
– Bruce Momjian’s Unlocking the Lock Manager, although you should probably have already seen his “MVCC Unmasked” talk, which I haven’t.
Check them out when they come to a conference near you.
I also learned a lot in the pub track, like I finally understand why TRUNCATE is so much faster than DELETE FROM table , and of course came up with silly ideas, like implementing query hints in the form of “warmer…warmer…colder…warmer…” and installing Pg on a router. Which, oddly, came up in two completely unrelated conversations. Must be a sign.
What’s more, I had a fine time in Chicago to boot – even got to do a little bike tour along the lake before my flight out on Saturday. I have next year’s dates on my calendar and will definitely be attending.
Something I’d like to see at future conferences: an “Ask the Expert” or “Pg Helpdesk” group discussion/hackfest/learning lab where people can get help with problems. I’ll be thinking about how to make that work, and suggest it for next year.
1 – DELETE FROM table deletes the individual rows; TRUNCATE removes the file from disk and just re-creates the table, and is therefore faster. Thanks, Aaron!