pgEdge Review (2026): Pricing, Features & Honest Verdict

TLDR

pgEdge is distributed PostgreSQL with active-active multi-master replication, built on standard Postgres instead of a fork. It is the right pick when you need multi-region writes without leaving the Postgres ecosystem behind. Best for: engineering teams running global apps that need local write latency and data residency control. Price: Free open source, commercial tiers custom priced. Rating: 7.9/10

What is pgEdge?

pgEdge is a distributed PostgreSQL platform. You run a cluster where every node accepts writes, and those nodes stay in sync through asynchronous logical replication. The replication engine is an extension called Spock. The part that matters most is what pgEdge did not do: there is no proprietary fork, no rewritten storage engine, no SQL dialect to relearn. Your existing extensions, drivers and ORMs keep working because underneath it is just Postgres.

The company was founded in August 2022 by Phillip Merrick and Denis Lussier. That pedigree is worth knowing. Lussier founded OpenSCG, which AWS acquired in 2018, after which he worked as a Postgres product manager on Aurora and RDS. Merrick co-founded EnterpriseDB. pgEdge came out of stealth with a $9 million round led by Sands Capital Ventures and Grotech Ventures, then raised a further $10 million led by Rally Ventures in 2024. Spock and the required Postgres patches are published on GitHub under the standard PostgreSQL License, so the runtime is genuinely open source rather than source-available with strings attached. We dug into the licensing because plenty of “open source” database vendors are not, and this one holds up.

What Are pgEdge’s Key Features?

Active-Active Multi-Master Replication

This is the headline capability and the reason pgEdge exists. Every node in a pgEdge cluster is writable. Nodes exchange changes asynchronously through Spock, with conflict detection and resolution built in rather than a consensus protocol sitting in the write path. This is a genuine fork in the road architecturally. CockroachDB and YugabyteDB both use Raft consensus, inherited from the Google Spanner design, which buys strict consistency at the cost of cross-region write coordination. pgEdge gives up some of that guarantee so writes complete at local speed in every region. If your workload is write-heavy across continents and can tolerate eventual convergence, this is the better shape. If you need serializable isolation across a global cluster, it is not, and no amount of marketing changes that trade. Spock handles the conflict cases with configurable resolution rules, but you still have to think about which tables can safely take concurrent writes in two places and which cannot.

Selective Table and Column Replication

pgEdge lets you choose what replicates where, down to individual tables and columns. In practice you can pin EU customer records to EU nodes while still replicating shared reference data everywhere. Most distributed databases treat data residency as a deployment concern, so you end up running one cluster per jurisdiction and stitching them together in application code. Handling it inside the replication configuration is cleaner and far less error prone. This is the feature we would actually buy pgEdge for if we were shipping into regulated markets.

ColdFront Hot and Cold Storage Tiers

ColdFront splits data across hot and cold storage while keeping both fully writable through the same SQL interface. pgEdge claims up to 90 percent lower storage costs on the cold tier. The real win is not the percentage, it is that you do not archive data out to a separate system and then maintain a second query path to reach it. For log tables, event histories and audit trails that grow forever and get read rarely, that saves both money and engineering time. Model it against your own access pattern before you build a budget on the headline number.

AI DBA Workbench and MCP Server

pgEdge has pushed hard into the agentic AI angle, and some of it is more than positioning. There is an MCP server so AI assistants can connect to the database through a standard protocol, RAG pipeline support for vector workloads, and an AI DBA Workbench that monitors and manages clusters continuously. Treat the AI DBA as an assistant rather than a replacement for someone who understands your query plans. The MCP server is the more immediately useful piece if your team is already building agents that need governed database access.

Managed Cloud, Kubernetes and BYOA Deployment

pgEdge Cloud runs managed clusters on AWS, Azure and Google Cloud with latency-based routing, automated backups, point-in-time recovery and a Terraform provider. The platform holds SOC 2 Type 2. The Enterprise Edition adds Bring Your Own Account, so nodes live inside your cloud account under your networking and compliance rules while pgEdge handles operations. Self-hosters get VM and Kubernetes editions instead. That spread of deployment models is wider than most competitors offer, and it is the practical reason a bank or a government agency can say yes.

How Much Does pgEdge Cost?

The open source core costs nothing. Spock, the Postgres patches and the published tooling sit on GitHub under the standard PostgreSQL License, and pgEdge states plainly that anyone is free to use, modify and deploy them independently at no cost. There is also a free tier on pgEdge Cloud, so you can stand up a managed cluster and evaluate the thing without talking to a salesperson first. That is more generous than most of this category.

Commercial pricing is where it gets frustrating. pgEdge does not publish rates for Distributed Postgres, Enterprise Postgres or the Cloud Enterprise Edition. Its own FAQ says VM Edition pricing is “competitive with other commercially supported PostgreSQL-based offerings” and then points you at sales. What the money buys is pre-built and tested binaries, system-level packaging and deployment tooling, security updates, and 24x7x365 support with defined response and resolution SLAs. Enterprise customers can add dedicated engineer time on top.

Compare that to the competition and the opacity stings. CockroachDB publishes real numbers: Basic is free until you cross 10 GB of storage and 50 million request units per month, Standard starts at $146 per month per two vCPUs, and Advanced starts at $295 per month per two vCPUs. Neon is cheaper again at small scale, with a free tier, Launch compute at $0.106 per compute unit hour and storage at $0.35 per GB-month. pgEdge asking you to book a call before you learn a price is real friction, and you should assume enterprise-tier contract sizes and negotiate accordingly.

Who is pgEdge Best For?

Use pgEdge if you are already committed to PostgreSQL and need to go multi-region without rewriting your data layer. It is a strong fit for:

• SaaS platforms serving users on multiple continents where write latency is hurting the product.
• Teams with hard data residency requirements that need per-table or per-column control, not just regional clusters.
• Organisations that want high availability and zero-downtime maintenance windows on Postgres without paying for a proprietary engine.
• Engineering teams that value keeping every Postgres extension, driver and tool they already depend on.

Skip pgEdge if you need strict serializable consistency across regions, because the asynchronous multi-master model does not give you that and pretending otherwise will cost you at 3am. Skip it if you are a solo founder or a small team running a single-region app, where managed Postgres from Neon, Supabase or your cloud provider is dramatically simpler and cheaper. And skip it if your procurement process cannot tolerate a vendor that will not publish a price. One more honest caveat: pgEdge is a small company against Cockroach Labs, EDB and the hyperscalers. The open source licence limits your downside because the code is genuinely yours to run, but support continuity is a fair question to raise in diligence.

Best pgEdge Alternatives

CockroachDB

The category leader for distributed SQL and the most direct competitor. CockroachDB uses Raft consensus for strict consistency, which is the opposite trade to pgEdge, and it publishes pricing: Basic is free to 10 GB and 50 million request units, Standard from $146 per month per two vCPUs, Advanced from $295 per month per two vCPUs. The catch is Postgres compatibility. Independent compatibility benchmarking scores CockroachDB well below native Postgres behaviour, so migrating an existing Postgres application is a real project. Choose CockroachDB if consistency guarantees matter more than compatibility.

YugabyteDB

The strongest open source option here. YugabyteDB is Apache 2.0 licensed, uses Raft like CockroachDB, and scores considerably higher than CockroachDB on Postgres compatibility because it reuses the Postgres query layer directly. Self-hosting is free, and YugabyteDB Aeon is the managed service if you do not want to run it. It is a better pick than pgEdge if you want distributed SQL with strong consistency and no licence cost, and a worse pick if active-active local writes are the whole reason you are shopping.

Neon

The cheap option, and honestly the right answer for a lot of teams who think they need distributed Postgres. Neon is serverless Postgres with a free tier, Launch compute at $0.106 per compute unit hour, Scale at $0.222 per compute unit hour, and storage at $0.35 per GB-month with no monthly minimum. It does branching, scales to zero, and costs a fraction of anything else on this list. What it does not do is multi-region active-active writes. If you are single-region and just want Postgres to stop being an operations problem, start here and only move to pgEdge when geography actually forces you to.

Final Verdict: Is pgEdge Worth It?

pgEdge earns a 7.9 because it solves a narrow problem unusually well. Multi-region active-active writes on genuinely standard PostgreSQL, with per-column residency control and a real open source licence, is a combination nobody else quite matches. The founders know this space from the inside, the deployment options run from bare metal to managed BYOA, and the free tier plus open source core means you can prove it out before committing budget. For a global SaaS platform hitting write latency walls, this is a serious tool.

The marks come off for two reasons. Pricing is hidden behind a sales call, which is a poor look next to CockroachDB and Neon publishing rates on a public page. And the asynchronous replication model is a real constraint, not a footnote: you are choosing availability and low latency over strict consistency, and you need to design your application around conflict resolution. Understand that trade before you sign. If it fits, pgEdge is excellent. If you need serializable global transactions or you are running one region, look elsewhere first.