London PostgreSQL meetup (29-07-25)
So how did you spend your tuesday evening?
Me, I spent the time hearing about different views of conversations I have had over the last couple of years/weeks/days. What do you mean I hear you ask? Well..
So I went to the meetup of London Postgresql group tonight, I will admit this was my first time and I was a bit unsure of what to expect.
What I found was the following:
- A lot of friendly faces that would like to say hello
- A free bar (ok this was not an incentive but hey it helped)
- Free food (ok you got me, my stomach is driving me here)
- Inspirational talkers that are passionate about the subjects that are discussing
To be honest the last point should be first, and not the last.
So what was talked about:
The age old battle of ids (ints/bigints) vs ids (uuids)
This was an in depth presentation of the differences between and integer or serial id and a UUID. I will admit as a long time supporter of int/bigint IDs over UUIDs I sort of got converted (I was thinking about the options anyway). I will not go deep into the conversation here (thats another blog post) but heres some of the things to think about:
- v7 uuids contain a time element - this means that yo can sort them better
- v4 random uuids and random INTs are similar in their performance profiles
- let the app designate the id, saves round trips and db latency
- v8 uuids allows you to specify the elements - just think sharding/client potential
PGROLL
We then had a talk from the creators of pgroll, the problem it was set to solve and how it solved this.
This proved quite insightful as in the past this has been trialed to see if it worked in our environment. After listening to the talk which was interesting and whilst possibly not something that would prove helpful for my current technical environment, its definately a good solution for others out there, some points taken from the talk were:
- Multi schema being visible to multi versions of application
- Controllable up and down migrations of databases in a yaml format
- Search path of the app (schema that the app looks at) altered depending on the version of the app your looking at
- Not so great in a multi schema environment
- Can only work in a solo migration env, so no concurrent migrations
- Not tested on generated columns that are persisted
Still even with these caveats definitely a powerful tool and possibly an essential tool in the old Postgresql tool belt.
So as I travel back on the train - slightly the worse for wear
I must say its definately worth going to user groups and I would advise everyone to go. This may not work in you locale but use the meetup site.
Lets be honest now, we like to think we know alot about the subject matter that we work on, but in reality theres alot of people out there that know alot more, we should learn from these people and not hide away from them (thats my view anyways).