Bucardo vs Gluesync

Bucardo is a free, open-source system for asynchronous PostgreSQL replication in multi-source and multi-target arrangements. It supports multi-way synchronization and one-direction master-standby replication. Users define named syncs that copy selected tables between a server and other servers. A Perl daemon listens for NOTIFY requests and connects to remote databases to copy data. Multi-source setups can use a standard conflict handler or a custom handler assigned per table. Configuration and replication databases require PostgreSQL 8.1 or later, while the configuration database also needs PL/pgSQL and PL/PerlU. The daemon needs a Unix-like operating system, is tested mainly on Linux and macOS, and does not run natively on Windows. Installation is available from source, RPM or Debian packages, and FreeBSD Ports. Bucardo does not replicate DDL or schema changes. The listed latest release is 5.6.0.

Gluesync replicates data changes in real time across relational and NoSQL databases, event streams and data lakes. Its log-based change data capture reads committed changes from transaction logs, journals or native CDC APIs instead of repeatedly polling full tables. The platform runs on customer infrastructure and supports on-premises, cloud, private-cloud and hybrid deployments. MOLO17 lists more than 31 supported platforms in four categories, with documented integrations including IBM Db2, PostgreSQL, Oracle, MySQL, Snowflake and Google BigQuery. The supported database list also includes systems such as Microsoft SQL Server, MariaDB, Cassandra, Couchbase and Amazon DynamoDB. An extensible API framework allows custom connectors. The web control plane includes visual pipeline orchestration, field functions, custom UDFs, live monitoring and SQL tools; monitoring covers throughput, latency and pipeline health, including from mobile. Listed security measures include TLS/HTTPS/WSS for data streams, AES-256 node-to-node encryption, OIDC, granular role-based access control, audit logging, and Kubernetes-native secrets and network policies. The maker says sensitive data is not persisted within the engine. Pricing is available on request, and a 30-day trial runs on the customer’s infrastructure with enterprise support during the trial. For a proof of concept, the trial page recommends a dedicated host with 8 vCPU, 8 GB RAM (16 GB recommended) and 250 GB disk.