Fork me on GitHub
#xtdb
<
2022-10-20
>
Mark Wardle08:10:33

Hi all. I am planning on switching to xtdb for a range of data storage uses, providing flexibility for different services to be composed and share the same backend data store, or separate stores, depending on the specific requirements for that deployment. I currently use a variety of stores directly such as lmdb, Lucene and SQL for different services. But I see xtdb core2 being discussed - does that mean it is now not advisable to use xtdb and development has switched to core2, and I should wait, or will there be easy migration, or do they serve different purposes (datalog vs SQL). When I read https://xtdb.com/core2/ it looks as if 1.x series is actually better suited to my needs.

tatut09:10:11

I am not an authoritative source, but afaict the XTDB 1.x series is still the one recommended for production

1
tatut09:10:37

as core2 is in research&development phase

tatut09:10:58

the page seems to confirm

Mark Wardle13:10:34

Thanks @U11SJ6Q0K - I've just spent some time reviewing the slack archives and see variants of this same question asked before so sorry for asking this.

Steven Deobald13:10:50

No apology necessary. 🙂 We actually have a few conversations queued up this week and next regarding the explanation of XTDB 1.x (vs) Core2. It's bound to come up quite a bit in the near term. That said, @U013CFKNP2R is there a way we could make the website clearer to folks like you who are considering 1.x in the near future?

Mark Wardle06:10:35

Thanks Steven. I don’t know but I think how prominent you make the core2 notice depends on lots of factors such as whether core2 is simply an exciting internal development / roadmap, is a new and separate parallel ‘thing’ that uses modern SQL and not datalog, or will shortly replace xtdb.

☝️ 1
tatut06:10:08

to avoid any osborne effect, that people will not use XTDB as they are waiting for the “next big version”

tatut06:10:42

the core2 stuff is pretty prominent in the frontpage

Steven Deobald12:10:10

Ah, right. That's fair. So, we recently attended Strange Loop and Core2 was the (bigger) focus of the two, since we wanted to get a feeling for a few things: • do people hate it? 😉 • is there as much appeal for a compute/storage separated db as we hear about within our own, smaller networks? • how does a broad audience feel about a columnar/HTAP immutable db? • where do peoples' problems lie on a spectrum of "immutable entities, plz" => "I have a very bitemporal problem and I want a shrinkwrapped bitemp solution" Unfortunately, shortly after the conference, we realized that http://xtdb.com (hooray for a memorable domain name) relegated the Core2 link to a little thing way down in the footer, where most people would never see it. Surfacing the Core2 stuff above the fold was to avoid confusion for those folks -- it's very good to know if it's had the opposite effect. (Even if the opposite effect itself is not a good thing.) 🙂 Thanks @U013CFKNP2R and @U11SJ6Q0K. I'll have a look at either taking that banner down, moving it, or figuring out a way to make it more obvious that Core2 is just an experiment for now. 🙏

roklenarcic13:11:14

Core 2 is moving toward columnar reporting/analytics DB?

refset14:11:55

The intention is for Core2 to be fast enough for regular transactional workloads at the same time. "HTAP" is the goal.