xtdb 2020-08-04 | Slack Archive

jonpither10:08:22

Hi all, we've published our video/blog on getting started with SQL / Crux using some simple steps - you may be interested: https://juxt.pro/blog/crux-sql

📺 3

🎉 6

ordnungswidrig13:08:24

Is it pronounced cru-se-cul?

ordnungswidrig13:08:54

Rather crucicle?

👍 3

devn23:08:24

I have a dumb question. How do I do a basic put submit-tx with a timestamp I’ve generated?

devn23:08:54

nevermind, user error

devn23:08:29

Is there any notion of temp-id like there is in Datomic?

devn23:08:42

or do I always need to explicitly create and specify the :crux.db/id ?

✔️ 3

refset23:08:25

Yep, all id usage must be explicit as the submit-tx API doesn't offer any id generation or temp id resolution or lookup-refs. However you can now at least write something to do the same job inside of your own transaction function 🙂

devn23:08:01

Thanks @taylor.jeremydavid. Pretty cool stuff. Looking forward to playing with it more.

🙏 3

devn23:08:14

One other question that I couldn’t find a direct answer to based on what I’ve read: What are the tradeoffs of not using Kafka? I’m wondering what I can trade in terms of operational complexity. The docs say: “When using Crux at scale it is recommended to use multiple Crux nodes connected via a Kafka cluster.” Some more information there would be nice!

devn23:08:52

For instance, how much scale do I get without bringing the operational complexity of a kafka cluster into the mix?

refset09:08:19

I haven't got any numbers to hand, but I would ballpark estimate that Kafka would be at least 10x faster under concurrent write load to execute submit-tx vs an equivalent active-active Postgres setup. You will also get better concurrent read performance with Kafka (for larger clusters of nodes) and I expect read latency to be lower, therefore end-to-end latency of Crux will be much lower in general. That said, I think crux-jdbc will still be more than fast enough for a large majority of workloads 🙂

devn23:08:02

Perhaps I’ve now found the answer to my own question: https://juxt.pro/blog/crux-jdbc

2020-08-04

Channels