Managed replication is still not backup
MSK Replicator is the easiest way to replicate between Amazon MSK clusters: fully managed, serverless, with consumer group offsets synced for you. But like every replicator, it copies corrupted records and accidental deletes just as faithfully as good data. Here is where each tool fits.
Two different jobs
AWS MSK Replicator continuously copies topics between Amazon MSK clusters — same region or cross-region — without any Connect infrastructure to run. AWS scales it for you and bills per GB replicated. It syncs topic configurations and consumer group offsets, and supports identical topic names across clusters, which makes failover straightforward. If you run MSK and need a warm standby region, it is a strong choice.
OSO Kafka Backup takes point-in-time backups of topic data, consumer group offsets, and topic configuration to object storage (S3, Azure Blob, GCS) or local disk, and restores any of it — to the same cluster, a new cluster, or a different platform entirely — with millisecond precision. The disaster recovery use cases cover the failure scenarios each approach handles.
The difference shows up on your worst day. When a producer bug poisons a topic, MSK Replicator delivers the poison to your standby cluster within seconds. A backup gives you the topic as it was before the incident.
| Feature | OSO Kafka Backup | MSK Replicator |
|---|---|---|
| Purpose | Backup / restore | Real-time replication |
| Point-in-time recovery | Yes | No |
| Protects against bad data / deletes | Yes | No |
| Works with | Any Apache Kafka | MSK on at least one side |
| Cross-account recovery | Yes | No |
| Storage backends | S3, Azure, GCS, local | Kafka only |
| Offset preservation | Built-in | Built-in (synced) |
| Infrastructure to manage | CLI or K8s operator | None (serverless) |
| Cost model | Object storage per GB stored | Per GB replicated + second cluster |
| Recovery time (RTO) | Minutes–hours | Instant (already replicated) |
| Data loss window (RPO) | Since last backup | Near-zero |
When to choose which
Choose OSO Kafka Backup when you need to recover from logical errors — bad deploys, accidental topic deletion, poison records — or compliance requires immutable copies outside the cluster. Backups also travel: restores can target a different account, region, cloud, or a self-managed cluster, with no requirement that MSK sits on either side.
Choose MSK Replicator when you run Amazon MSK and need low-RPO regional failover without operating MirrorMaker 2 yourself. Offset sync and identical topic names make consumer cutover clean, and there is no replication infrastructure to patch or scale.
Run both when MSK is production-critical: MSK Replicator for availability during regional failure, scheduled backups for recoverability from data-level incidents. Replication and backup protect against different failure modes — the full feature comparison walks through the same distinction for MirrorMaker 2 and tiered storage.
Frequently asked questions
Does AWS MSK Replicator support point-in-time recovery?
No. MSK Replicator performs continuous forward replication between clusters. It cannot restore a topic to an earlier state — if bad data reaches the source, it is replicated to the target within seconds. Point-in-time restore requires a backup taken before the incident.
Can MSK Replicator copy data to S3 or another storage service?
No. MSK Replicator replicates from Kafka to Kafka, with an MSK cluster on at least one side. OSO Kafka Backup writes compressed topic data, offsets, and topic configuration to S3, Azure Blob, GCS, or local disk, independent of any cluster.
Does MSK Replicator preserve consumer group offsets?
Yes — offset synchronization is built into MSK Replicator, so consumers can fail over to the target cluster near their last position. OSO Kafka Backup also preserves consumer group offsets, but in backups: restores bring offsets back consistent with the restored records, at the restore point you choose.
Is OSO Kafka Backup cheaper than MSK Replicator?
For data protection, usually. MSK Replicator bills per GB replicated plus inter-region transfer, and requires a second MSK cluster running continuously. Backups are billed as object storage per GB stored, with Zstd or LZ4 compression reducing the footprint. For low-RPO availability, replication is the right spend — the two solve different problems.
Can I restore an MSK backup to a non-MSK cluster?
Yes. OSO Kafka Backup works with any Apache Kafka distribution, so a backup taken from MSK can be restored to self-managed Kafka, Strimzi on Kubernetes, or another cloud provider — useful for migrations and multi-cloud DR that MSK Replicator does not cover.
Ready to protect your Kafka data?
Take your first backup in minutes with the open source CLI, or talk to us about Enterprise features like encryption, RBAC, and audit logging.