AWS announced on November 10, 2026 that Amazon DocumentDB (with MongoDB compatibility) 8.0.3 is now available in every Region that runs the service. The engine update adds hybrid post-quantum key exchange for encrypted connections, wildcard indexes, the $unionWith aggregation stage and a handful of smaller MongoDB API gaps that are now closed.
The release matters for two groups. Security teams get protection against attackers who capture traffic today and hope to decrypt it with a future quantum computer. Developers moving MongoDB 8.0 workloads get more query features that previously forced code changes or workarounds.
What Is New in DocumentDB 8.0.3
The AWS announcement and the engine release notes list these changes for version 8.0.3 (engine version 4.0.13572):
- Hybrid post-quantum key exchange for TLS 1.3, based on ML-KEM combined with a classical elliptic curve exchange.
- Two CloudWatch metrics, TLSHandshakesPostQuantum and TLSHandshakesClassical, that count successful handshakes of each kind per instance.
- Wildcard indexes that cover all fields in a document, or all fields below a given path.
- The $unionWith stage, which merges results from several collections inside a single pipeline.
- $lookup and $graphLookup inside views built with createView, plus $unionWith and $facet inside correlated $lookup sub-pipelines.
- Change stream events of type dropIndexes when dropIndex() or dropIndexes() succeeds.
- Per-collection compression figures in collStats and stats(), including input bytes, output bytes, operation count and ratio.
Two bugs are also fixed: a wrong retryable write error label from findAndModify, and a sort problem when the query planner uses disjoint index scans.
How Post-Quantum TLS Works on DocumentDB
The new key exchange runs on instance-based clusters and is switched on by default for every value of the tls cluster parameter except disabled. No configuration change is needed, and certificates, cipher suites and allowed protocol versions are untouched. Clusters on engine versions older than 8.0.3 keep using classical key exchange only.
The cluster offers three hybrid groups, in order of preference: X25519MLKEM768, SecP256r1MLKEM768 and SecP384r1MLKEM1024. A connection only goes post-quantum when the client speaks TLS 1.3 through a library that understands ML-KEM. Everything else falls back to the classical exchange, so older apps keep working.
The AWS documentation lists minimum client versions per language:
- C, C++, PHP and Ruby drivers linked against OpenSSL 3.5 or newer.
- Go 1.24 or newer, and Node.js 24.5 or newer.
- Java with JDK 27 or newer, or with the Bouncy Castle JSSE provider.
- .NET running on Linux, and Python builds, when either uses OpenSSL 3.5 or newer.
- Rust, when the mongodb crate is compiled with its rustls-tls-aws-lc feature.
Watch Out for Firewalls and Handshake Retries
A client hello that offers ML-KEM weighs roughly 1.2 to 1.5 KB, which is bigger than a classical one and can be split across packets. AWS warns that some aging firewalls and proxies that inspect encrypted traffic may drop these sessions. If an app breaks after the upgrade but connects fine with classical groups only, suspect the network gear first.
There is also a small latency trap. If an ML-KEM capable driver leaves the matching key share out of its first message, the cluster sends it back for a second attempt. The session still ends up post-quantum, but the added round trip makes setup about as slow as TLS 1.2, so make ML-KEM groups the first choice in the client.
To confirm a single connection, AWS shows an openssl s_client test with OpenSSL 3.5 and the DocumentDB CA bundle, offering only X25519MLKEM768. A successful run reports that group as the negotiated TLS 1.3 group.
What It Means for Developers
For most teams the post-quantum change is free protection that arrives with a routine engine upgrade. The real work is on the client side, since a Java 21 service or a Python build tied to an older OpenSSL will quietly stay on the classical exchange. The new CloudWatch metrics show what share of traffic is protected.
Wildcard indexes are useful for collections with dynamic or user-defined attributes, such as product catalogs or event payloads, where you cannot predict every field that will be queried. They are not a replacement for well-chosen compound indexes on hot query paths, so test both the query plans and the write overhead before relying on them.
The $unionWith stage and $lookup support in views remove common blockers when moving MongoDB aggregation code. Reports that combine archive and live collections, or views that join reference data, can now run on DocumentDB as written. The dropIndexes change stream event also helps teams that sync schema changes to search indexes or caches.
How to Upgrade
Version 8.0.3 applies to DocumentDB clusters on the 8.0 engine line. Plan the upgrade in a test cluster first, then confirm that every client connects through any firewalls, load balancers or inspection proxies in your network path. Watch the two new handshake metrics after the change to see which services still negotiate the classical exchange.
If you run DocumentDB Serverless, our recent look at DocumentDB Serverless platform version 4 covers the latest throughput gains on that side of the service.
Conclusion
DocumentDB 8.0.3 is a solid compatibility and security release. Post-quantum TLS turns on by itself once clients are recent enough, while wildcard indexes and $unionWith make it easier to move MongoDB 8.0 applications without rewriting queries. The main thing to test is whether your network gear handles the larger TLS handshake.
Sources
- AWS What's New: Amazon DocumentDB now supports post-quantum TLS, wildcard indexes, and $unionWith in version 8.0.3
- Amazon DocumentDB Developer Guide: Release notes
- Amazon DocumentDB Developer Guide: Encrypting data in transit and post-quantum key exchange
See my articles more often when you search Google.
