Neki update: $50 Metal, read-only replicas, Azure, and more

Neki, our sharded Postgres, has been in platform preview for about a month. The team continues to work towards the path to production-readiness, and we've added quite a few new features and capabilities since launch.

Here's a quick look at what's landed since the preview opened and how you can start evaluating Neki for your workload today.

What's shipped

We've recently added both read-only replicas and smaller Metal instances to Neki starting at just $50 per month.

Read-only replicas were a hotly-requested feature for Postgres, particularly for customers with a large, global userbase. Neki now also supports read-only replicas for shards in your cluster.

We also now support smaller Metal node sizes, going all the way down to database nodes with 1/8th vCPU, 1GB RAM, and 10GB of locally-attached NVMe SSD.

You can try both of these capabilities on Neki in PlanetScale today. These are just a few amongst many others, including:

• Manage Neki extensions with Terraform and the CLI (Sept 18). Enable extensions by name with extensions in a planetscale_neki_configuration_profile resource or with pscale branch config-profile update --extensions, and change their settings in the same update.
• Branch metrics in the PlanetScale MCP server (Oct 9). planetscale_get_branch_metrics pulls CPU, memory, disk, replication lag, and connections into your AI agent, or wherever you connect to your infrastructure via MCP. It works for Neki, Postgres, and Vitess.
• PlanetScale on Azure (Oct 13). You can now create Neki and Postgres databases in Azure regions, alongside AWS and Google Cloud.
• Neki router parameters in Terraform (Oct 14). Terraform provider v1.13.0 adds parameter management to planetscale_neki_router, and the planetscale_databases data source can now filter by kind (mysql, postgresql, or neki).
• CLI shows which parameters need an update (Oct 14). CLI v0.346.0 adds a "requires update" column for Neki configuration profiles, admins, and sidecars, plus a requires_update field in JSON output.
• Huge pages for Neki and Postgres (Oct 15). Set huge_pages to on in a configuration profile's Parameters tab on supported cluster sizes. The change requires a restart.

Under the hood: query layer

Much of the engineering effort since preview has gone into the query layer: making more of your existing SQL run on a sharded database without changes, and making the standard Postgres tools you already use work against Neki.

The team has merged more than 650 changes since the preview opened. A few more highlights:

• More of your SQL just works. INSERT ... SELECT ... ON CONFLICT, cross-shard UPDATE ... FROM, DELETE ... USING and MERGE, correlated EXISTS almost anywhere in a query, ARRAY(SELECT ...), SQL cursors, LISTEN/NOTIFY, materialized views, and PL/pgSQL exception handlers are all now supported.
• Export with standard tools. COPY ... TO STDOUT now works on sharded tables in text, CSV, and binary, so pg_dump and other export tools can read data out of a sharded Neki database.
• Faster bulk writes. Distributed INSERT ... SELECT is about 76x faster than at launch, and COPY ingest is roughly 5x faster.
• See how your query runs on every shard. EXPLAIN (NEKI_PG_PLAN) shows Neki's distributed plan together with the Postgres plan each shard ran. The output follows Postgres's own EXPLAIN format, so pgAdmin, DBeaver, explain.dalibo.com, and explain.depesz.com can read it.

Start evaluating now

If you're running Postgres and expect to need sharding or any other Neki features like online schema changes or no downtime version upgrades, now is a good time to start the conversation. Email us at [email protected] or fill out the form, and we'll set up a call with our solutions team.

If you want to learn more about Neki, we recommend starting with When to shard, Neki best practices, and Data topology.