Dgraph Release v23.1.1 is now Generally Available

Dgraph 23.1.1 is now generally available!

In addition to a few tooling optimizations, this release dramatically improves performance for queries where a small number of items is being filtered by a large type index by avoiding a large in-memory operation (in extreme cases this increases query speed by 100x).

These optimizations are in preparation for Dgraph v24 with Vector support (currently in Alpha). If you’d like to try v24-alpha on Dgraph Cloud please email vector@hypermode.com.

Please file Github Issues with the label v23.1.1 if you find any problems or bugs in this release.

6 Likes

In the Pull Request for the performance improvement, the following is given as an example:
func(uid: 0x1) @filter(type)

Is this implemented for filters in general? or only for ‘type’ filters?

We use UID lookups with filters extensively.

Thanks,
Ryan

Right now, its only for the type filters. But we have plans to improve more filters() using this.

2 Likes

We use a lot of this style in upsert/queries:

node as var(func:uid("0x1"))
nodeDate as var(func:uid(node)) @filter(has(publication.date))
nodeDateFiltered as var(func:uid(nodeDate)) @filter(le(publication.date, "2023-10-05T10:00:00Z"))

Love to see any improvement on the has and le filters.

Thanks,
Ryan

1 Like

is 23.1.1 v is a stable version? and can be upgraded in production?

Yes, 23.1.1 is stable version. It can be directly replaced with the 23.1 binary.

We have successfully moved to it. Very easy drop-in replacement, no issues.