Percona ClusterSync 1.0.0 Adds High Availability and Sharding
Percona announced on October 6, 2026, that Percona ClusterSync for the open-source document database, or PCSM, has reached version 1.0.0 and is now generally available. The tool is described as production-ready, fully supported, and already proven on real multi-terabyte migrations. The announcement came from Radoslaw Szulgo, who wrote on behalf of the product and engineering team that PCSM helps teams move off the cloud-hosted version, the enterprise edition, or the community edition of the document database to Percona Server for the document database, or move a cluster between data centers, with near-zero downtime.
PCSM is a cluster-to-cluster synchronization tool. In plain terms, a cluster is a group of database servers that work together as one system. The tool replicates data from a source document database cluster to a target cluster while applications continue to run. It first clones the existing data, then streams every change in real time until the team is ready to cut over. PCSM automatically handles all indexes and Data Description Language (DDL) operations, which are commands that define or change the structure of a database, such as creating an index or altering a collection. The release also supports filtered synchronization, so a team can choose to copy only a specific database or collection instead of the entire cluster. Percona positions this as a solution for leaving vendor lock-in with the cloud-hosted document database service, migrating a deployment to a different data center, or troubleshooting production data in an isolated, cloned environment.
The biggest addition in version 1.0.0 is high availability for the replication process itself. A team can run several PCSM instances with the same source and target. One instance actively replicates while the others wait on standby. If the active instance fails, a standby automatically takes over and resumes from the last checkpoint, which is a saved position in the replication stream. Because instances can be placed across different data centers, a failed node or even a failed site no longer halts a migration. Sharding support is now generally available as well. Sharding is a technique that splits a large database across multiple servers so it can handle more data and traffic. PCSM can migrate from a sharded source to either a sharded cluster or a replica set. For sharded-to-sharded moves that use ranged shard keys, PCSM pre-splits target chunks to match the source boundaries, which reduces work during the final switch. PCSM 1.0.0 also adds parallel sync to multiple targets, meaning one source cluster can replicate to several target clusters at the same time. Each PCSM instance has its own namespace filters, so every target can receive a different set of data. Finally, the release extends platform support with new packages for RHEL 10, Rocky Linux 10, AlmaLinux 10, a widely used enterprise Linux distribution version 10, and Debian 13 (Trixie), on both x86_64 and ARM64 architectures. The release also improves error handling, connection pool tuning, and Kubernetes compatibility. Percona points to the 1.0.0 release notes for the full list of changes and upgrade steps.
Percona says PCSM has already carried real production workloads off the cloud-hosted document database service. PriceShape, a Danish pricing-intelligence software-as-a-service company, had run out of cloud service disk capacity and hit oplog limits on its update-heavy workload. An oplog is a record of every change made to a document database, and when it fills up, replication can stall. With PCSM, PriceShape moved about 6.6 terabytes of data from the cloud-hosted document database service to Percona Server for the document database running on Amazon EKS, a managed Kubernetes service. PCSM performed the initial full sync and then kept both clusters in lockstep via change-stream replication, which is a real-time feed of changes, until cutover. The switch finished inside the company's required downtime window, with no impact on users. Today PriceShape runs without the old disk and oplog constraints and is planning to shard for further growth. Mads Toustrup-Lønne, VP of Technology at PriceShape, said in the announcement: 'With Percona, we can always get the help and support we need with just a Slack message, and the support engineers take their time reviewing our environment before answering, instead of just referring to the documentation. We knew the migration wouldn't be without its hiccups, but Percona quickly involved the relevant persons to help us move forward as soon as something came up, and we completed the migration without impacting our customers.'
For many teams, moving a production database is one of the riskiest operational projects. The combination of an initial clone, continuous change streaming, and automatic failover during replication means a migration can be tested and rehearsed before the final switch, and the final switch can be kept short. The high availability feature in PCSM 1.0.0 is a transferable lesson: any critical data movement should itself be protected against infrastructure failure, not just the database it moves. For teams that run database migrations or manage multiple clusters, using a managed cloud and data infrastructure provider such as AEU Data can reduce operational risk. AEU Data offers managed databases, automated backups, encrypted storage, monitoring, and high availability, so the target side of a migration is protected while the sync runs.
PCSM 1.0.0 is free and open-source, backed by Percona Support for teams that want expert help during a migration. Percona provides installation instructions, an explanation of how PCSM works, a list of supported document database deployments, and the PriceShape customer story. Teams planning a move off the cloud-hosted service or between data centers can talk to Percona for planning help.