The choice is not really about support quality. It is about whether you want your support arrangement to also decide your distribution.
Confluent supports Confluent Platform. Independent providers support Apache Kafka, wherever it runs. Both can be excellent. What differs is what each one assumes about where your architecture is heading.
What Each One Actually Covers
Confluent support covers Confluent Platform: its distribution of Kafka, plus Schema Registry, ksqlDB, its connector catalogue, RBAC, tiered storage and control plane tooling. It bundles licensing for those components with the support service, so the invoice covers software and support together. The advantage is single-vendor accountability — the people supporting the code are the people who wrote it.
Independent support covers Apache Kafka itself: brokers, KRaft or ZooKeeper, topics, partitions, replication, consumer groups, producers, and the operational reality of running them. It is a service against software you already run at no licence cost, and it does not assume anything about your distribution — including whether you keep it.
Where Lock-In Actually Appears
Not in Kafka. Kafka is Apache-licensed, portable, and running it does not tie you to anyone.
Lock-in appears in the layer above it. Schema Registry, once your producers and consumers depend on it. ksqlDB queries, once stream processing logic lives in them. Confluent-specific connectors. RBAC and tiered storage implementations. Control plane tooling your operations team has built habits around.
None of that is a criticism — those components exist because they solve real problems. The point is that each one you adopt raises the cost of ever changing provider, and that cost is much easier to evaluate before you build on them than after.
The Decision, In One Question
What proportion of your architecture depends on proprietary components?
Heavy dependence — Schema Registry central to your contracts, real stream processing in ksqlDB, a fleet of Confluent connectors — points to staying with the vendor. Replacing those is a project, not a procurement decision.
Vanilla Kafka — open source brokers, your own tooling, standard clients — points to independent support. You are paying vendor prices for a bundle whose bundled half you do not use.
Somewhere in between is where most estates actually sit, and the useful exercise is boring: list every proprietary component in use, and for each one write down what would replace it. That list is the real decision. Anything shorter than you expected is a strong signal.
What Actually Breaks in Production
One observation that shifts this comparison more than the feature matrices do.
Kafka source defects are rare. What causes real production incidents is partitioning that no longer fits the workload, consumer groups rebalancing under load, producer timeouts under back-pressure, retention and disk pressure, and capacity that was sized for last year. Those are operational problems, and they are solved by people who know Kafka and know your cluster — not by proximity to the committers.
That is why the vendor's strongest argument — accountability for internal defects — applies to a smaller share of incidents than it appears to. Related reading: diagnosing consumer lag, rebalance storms and evaluating a Kafka support SLA.
AceMQ provides independent Apache Kafka support for production estates. If you want to compare what you currently pay against what you actually use, get in touch.
FAQ
What is the actual difference between Confluent support and independent support?
Confluent supports its own distribution and proprietary components. Independent providers support Apache Kafka wherever you run it, without presuming anything about your distribution.
Do I need Confluent Platform to get good Kafka support?
No. Apache Kafka is production-grade on its own, and large estates run it with independent support and no Confluent components.
Where does lock-in actually appear?
In the proprietary layer — Schema Registry, ksqlDB, vendor connectors, RBAC and tiered storage — not in Kafka, which is open source and portable.
Is independent support cheaper?
Usually, but the comparison is not like for like: vendor pricing bundles platform licensing with support. Whether that half is worth it depends on whether you use it.
What do you lose by going independent?
The proprietary components, and single-vendor accountability for internal defects — which matters less than it sounds, because most incidents are operational rather than source defects.
How should I decide?
List every proprietary component you actually depend on and what would replace each. That list is the decision.
