How scalable is Camunda for large-scale enterprise applications?

Camunda 7 Topics
1

Hi all,

Can anyone suggest How scalable is Camunda for large-scale enterprise applications?

Thanks in Advance

Kamal Hinduja

1 Like
2

Hi @kamalhinduja,

Interesting question, but probably too broad to answer.

There are plenty of ways how you can scale Camunda. E.g. if most of the work is done by task workers then you can just create more task worker instances. It’s basically the same for micro processes with embedded engine, those can also be implemented as workers on a message queue and you can create more instances on demand.

So it really depends on your needs and how much time and resources you want to invest to make the application highly scalable.

3

Hi @kamalhinduja ,

Camunda 7 and Camunda 8 differ significantly in architecture and scalability. I’ll compare both carefully:

:one:Camunda 7 Scalability

➔ Architecture

  • Monolithic architecture (though can be clustered).
  • Typically runs as part of a Java Application (embedded engine) or as standalone server (shared engine).
  • Uses RDBMS (PostgreSQL, Oracle, etc.) as central state store.
  • Process state, history, and job data all live in relational DB.

➔ Strengths

  • Mature, stable, highly battle-tested.
  • Can handle tens of thousands of concurrent process instances if properly tuned.
  • Database clustering, application server clustering (e.g., WildFly, Tomcat).
  • Horizontal scaling possible (shared DB, multiple engine nodes).
  • Sophisticated job executor tuning.

➔ Challenges at very large scale (> millions of instances)

  • Database becomes the bottleneck.
  • DB contention on writes (especially for jobs, history, and execution state).
  • Complex failover/recovery as DB lock management can impact uptime.
  • Heavy history data may require separate archiving strategy.
  • Async continuations & external task workers scale well, but long-running processes need careful design.

➔ Typical sweet spot:

  • 50K - 500K concurrent active process instances with good DB design.
  • Large batch processing is doable, but requires database tuning.
  • Excellent for enterprise apps with moderate to high complexity, but not hyperscale cloud-native workloads.

:two:Camunda 8 Scalability

➔ Architecture

  • Cloud-native microservice architecture (designed for scale).
  • Zeebe workflow engine (event sourcing based).
  • State stored in RocksDB on local disk (no RDBMS for runtime state).
  • ElasticSearch for operational data.
  • Kafka-like partitioned model.

➔ Strengths

  • Built for massive horizontal scaling.
  • Partitioned, distributed event log allows millions of workflow instances.
  • Much higher throughput than Camunda 7.
  • External task workers are easier to scale (gRPC-based job workers).
  • ElasticSearch decouples history queries from workflow state.
  • Suitable for real-time, IoT, telecom, finance, etc.

➔ Challenges

  • Zeebe is eventually consistent — design mindset must change.
  • Requires operational knowledge of Kubernetes, ElasticSearch, S3 backups, etc.
  • Partition sizing requires capacity planning.
  • Complex backup & restore procedures.
  • Maturity of some advanced BPMN features is still catching up with Camunda 7 (though improving fast).

➔ Typical sweet spot:

  • 500K to 50M+ active process instances depending on partition count.
  • Linear scaling across nodes.
  • Very well-suited for hyperscale workloads, microservices orchestration, and streaming pipelines.

:microscope:Quick summary table

Feature Camunda 7 Camunda 8
Architecture Monolithic + DB Cloud-native, distributed
Scalability Limit DB bound, 100K-500K Horizontally scalable, 50M+
Database RDBMS (Postgres, Oracle) RocksDB + ElasticSearch
Job execution Thread-based Distributed workers
Best for Traditional enterprise apps Cloud-native, high volume
Deployment Java app server Kubernetes, SaaS or Self-managed
State consistency Strong consistency Eventual consistency

:three:Enterprise-level Scalability Use Cases

Use case Camunda 7 Camunda 8
Banking core systems :white_check_mark: :white_check_mark:
Telecom IoT event handling :warning: :white_check_mark::white_check_mark:
Insurance claim processing :white_check_mark: :white_check_mark:
High volume API orchestration :warning: :white_check_mark::white_check_mark:
Complex human workflows :white_check_mark::white_check_mark: :white_check_mark:
Streaming event orchestration :x: :white_check_mark::white_check_mark:

:wrench: Rule of thumb recommendation

  • If you’re building traditional enterprise apps → Camunda 7 is still great, mature, stable, lots of plugins.
  • If you need hyperscale, cloud-native orchestration → Camunda 8 is the future-proof choice.
1 Like