Oct 11, 2026 · 11 min · 6 segments
In this episode of Scaling Postgres, we discuss performance improvements in ParadeDB as a result of PlanetScale, an open source BM25 index from PGX Inc., pgbouncer and pgvector patching and direct I/O…
Creston JamisonHost
Our first piece of content, PlanetScale release Text Search, and we have a lot to say, part one.

So paradb.com, if you're not familiar, they have actually embedded Tantivy to give Postgres BM25 ranked text search in Postgres.

And I did cover this article a number of weeks ago, and this is what the PlanetScale article looks like.

They released Ten, which uses CTIDs to help track and do text searching, and you can see the really high performance difference between ParadeDB and Ten, about tenfold, at least with the dataset and the queries that they were using.

And I particularly like on the graph here, you can show balls, which basically shows how much faster Ten is compared to ParadeDB, and heaven help you if, if you're using Postgres gen for doing text searching.

But upon seeing this, and they say in this blog post, you know, quote, "Ten is fast, at least eight times faster than ParadeDB 0.25 in every PlanetScale benchmark." And the quote continues, "So fast that the only response which made sense was to shut up and put on our performance optimization hats." So kudos to them.

And they tried to replicate the benchmarks using the same Stack Exchange benchmark dataset, harness, and machine types.

And with the optimization they've done, they do see in certain cases, like looking at mixed queries and disjunction queries, maybe now it looks like they're three times faster than Ten.

Now, are they truly three times faster? Again, you always have to take benchmarks with a grain of salt.

But what I do appreciate is they saw how fast Ten was on PlanetScale, so they rolled up their sleeves and said, "All right, how can we make this faster?" And at least according to these benchmarks, clearly they have.

Now, one reason that they say that PlanetScale says Ten is faster is it identifies rows by CTID values, whereas Tantivy has to map values from doc ID to CTID values.

So basically, this map makes things slower compared to just using CTID values as the direct document identifiers.

But they question that, and some of the changes they made, they're still using the Tantivy engine, which uses that, so they're still mapping the Tantivy doc IDs to CTIDs.

And they looked at another Hacker News, and it appears that ParadeDB is twice as fast or more across various different metrics, although not as, quite as much with two terms in a top K search.

So in terms of what they did, they also saw a few cases where it kind of depends on how things were configured.


But in terms of what ParadeDB did is they started writing a denser structure with better data locality for finding documents.

And third, a few other small optimizations related to lazy reading of other pieces of data.

But they said most of these changes were happening in the Tantivy library, apparently not in their integration code.

Our first piece of content, PlanetScale release Text Search, and we have a lot to say, part one.

So paradb.com, if you're not familiar, they have actually embedded Tantivy to give Postgres BM25 ranked text search in Postgres.

And I did cover this article a number of weeks ago, and this is what the PlanetScale article looks like.

They released Ten, which uses CTIDs to help track and do text searching, and you can see the really high performance difference between ParadeDB and Ten, about tenfold, at least with the dataset and the queries that they were using.

And I particularly like on the graph here, you can show balls, which basically shows how much faster Ten is compared to ParadeDB, and heaven help you if, if you're using Postgres gen for doing text searching.

But upon seeing this, and they say in this blog post, you know, quote, "Ten is fast, at least eight times faster than ParadeDB 0.25 in every PlanetScale benchmark." And the quote continues, "So fast that the only response which made sense was to shut up and put on our performance optimization hats." So kudos to them.

And they tried to replicate the benchmarks using the same Stack Exchange benchmark dataset, harness, and machine types.

And with the optimization they've done, they do see in certain cases, like looking at mixed queries and disjunction queries, maybe now it looks like they're three times faster than Ten.

Now, are they truly three times faster? Again, you always have to take benchmarks with a grain of salt.

But what I do appreciate is they saw how fast Ten was on PlanetScale, so they rolled up their sleeves and said, "All right, how can we make this faster?" And at least according to these benchmarks, clearly they have.

Now, one reason that they say that PlanetScale says Ten is faster is it identifies rows by CTID values, whereas Tantivy has to map values from doc ID to CTID values.

So basically, this map makes things slower compared to just using CTID values as the direct document identifiers.

But they question that, and some of the changes they made, they're still using the Tantivy engine, which uses that, so they're still mapping the Tantivy doc IDs to CTIDs.

And they looked at another Hacker News, and it appears that ParadeDB is twice as fast or more across various different metrics, although not as, quite as much with two terms in a top K search.

So in terms of what they did, they also saw a few cases where it kind of depends on how things were configured.


But in terms of what ParadeDB did is they started writing a denser structure with better data locality for finding documents.

And third, a few other small optimizations related to lazy reading of other pieces of data.

But they said most of these changes were happening in the Tantivy library, apparently not in their integration code.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.