dsync vs Gluesync

dsync moves data between databases and other data stores through one-time migrations or ongoing, one-way replication. After the initial copy, it can watch the source and apply later changes at the destination. The open source edition is a single binary that can run on a laptop, VM, or Docker container. Connectors include MongoDB, Cosmos DB, DynamoDB, AWS DocumentDB, PostgreSQL, S3, and CSV files; several others are still in preview or development. Checks range from quick namespace counts to full hash-based verification, and saved state lets an initial copy or change-capture flow resume. Progress appears in the CLI and a web page hosted on the machine running dsync. Adiom says dsync does not store data and supports network encryption. Transformations can use external processes connected over gRPC. It does not replicate source DDL statements or indexes. With a metadata store for resumability, the open source version has a 16 MB read-plan cap. Enterprise adds horizontal scalability, observability, and transformations. The open source plan is free.

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.