Speaker Interview: Tomas Vondra

PostgreSQL Performance Tuning   Tuesday 09:00   Berlin A

CREATE STATISTICS - what is it for?   Thursday 10:30   New York

Twitter: fuzzycz LinkedIn: tomasvondra Company website: 2ndquadrant.com Company Blog: blog.2ndquadrant.com Google+: Tomas Vondra Reddit: cats

Could you briefly introduce yourself?

I live in Prague, and I'm a PostgreSQL developer, consultant, support engineer, trainer and all that. I do work for 2ndQuadrant, one of the companies providing various PostgreSQL-related services. So I'm quite lucky, because my current job intersects with my hobby.

How do you engage with the PostgreSQL Community?

I do participate in the development efforts - mailing list discussions, contributing patches and reviewing patches submitted by others, and now also committing them. Aside from that, I'm one of the people running CSPUG (Czech and Slovak PostgreSQL Users Group), which also means I'm organizing the yearly Prague PostgreSQL Developer Day conference. And of course, I'm attending various other conferences, either to give a talk or just as a regular attendee.

Have you enjoyed previous pgconf.eu or FOSDEM conferences, either as attendee or as speaker?

Of course, that's why I'm coming back ;-) I've missed one or two pgconf.eu conferences. It's a pretty unique experience, unlike most other conferences (particularly non-PostgreSQL ones) that I've attended over the years.

What will your talk be about, exactly? Why this topic?

I'll be talking about query planning issues caused by correlated columns, and how multi-column statistics (CREATE STATISTICS introduced in PostgreSQL 10) can help with that. I'm giving this talk because it's a fairly common cause for slow queries, and it's useful why we have this issue and what tools we have to deal with it - either in the current version, or in the future.

What is the audience for your talk?

Anyone who has to deal with slow queries - analyze query plans to understand why the query is slow and how to fix it. Often creating an index is enough, but sometimes the actual problem are poor estimates - and that's what this talk is about.

What existing knowledge should the attendee have?

A basic understanding of query planning and cardinality estimation would be helpful, that's for sure. But it's not a requirement - I'll briefly explain the basics in the talk itself.

What is the one feature in PostgreSQL 11 which you like most?

That's a tough one. There are so many significant improvements to various parts, including parallelism and partitioning, etc. But if I had to name just one discrete feature, I'd say it's either the CREATE INDEX parallelism, or ALTER TABLE … ADD COLUMN with a default (without having to rewrite the whole table).

Which other talk at this year’s conference would you like to see?

D'oh, it's so difficult to pick one, because there are so many interesting talks and it also depends on the attendee (DBA vs. app developer vs. PostgreSQL developer vs ….).

As a PostgreSQL developer, I certainly want to see talks by fellow developers - Heikki Linnakangas, Andres Freund, Amit Kapila and other "usual suspects". I always learn something new from them, either something small and immediately useful during development, or different point of view on a problem. If I had to pick one, I'd say "Pluggable Storage in PostgreSQL".

As a user/DBA/app developer, I'd like to see "Partition a 10TB table" and "Use Logical Decoding to build your own application cache" and a couple others. I expect those to be talks with plenty of immediately useful information.

But that's the thing - don't go just to talks talking about topics you like and are already familiar with. Go to talks you did not really plan to attend, and I'd bet you won't regret it.