Camunda 8.8 Zeebe exportless Architecture

Camunda 8 Topics
,
1

Hi Team,

I would like to know detail about this. Zeebe we use elastic exporter to export data into ES db until 8.7 . So what’s new in this 8.8 about Zeebe exportless architecture. .Because I read that directly Zeebe streaming into operate which means operate will not query from ES ? Can you please give clarify me

2

Hi @vinothkumar,

Great question about the new exportless architecture in Camunda 8.8! Let me clarify what has changed and how it affects the data flow.

What’s New in Camunda 8.8: Exportless Architecture

The “exportless architecture” in Camunda 8.8 refers to a significant change in how data flows from Zeebe to Operate and other components:

Before 8.8 (Until 8.7):

  • Zeebe used the Elasticsearch/OpenSearch exporter to write data to ES/OS indices
  • Operate had separate importers that would read from these Zeebe indices and import the data into Operate-specific indices
  • This created a multi-step process: Zeebe → ES indices → Operate importer → Operate indices

In 8.8 and Later:

  • Zeebe now uses a unified Camunda Exporter that writes directly to the “Camunda indexes” (the secondary storage schema used by Operate, Tasklist, and Identity)
  • No separate Operate/Tasklist importers needed for new data - Zeebe streams directly to the target indices
  • This removes the importers from the “hot path” and simplifies the data flow

Key Benefits:

  • Fewer moving parts - no Operate/Tasklist importer to manage
  • Lower latency - data appears faster in Operate/Tasklist
  • Fewer synchronization issues between components
  • Simplified operations and troubleshooting

What About the Old Elasticsearch/OpenSearch Exporters?

The ES/OS exporters are still available and fully supported in 8.8+! They remain useful for:

  • Optimize (which still needs them)
  • Custom use cases and integrations
  • Migration scenarios

However, they are no longer the primary source for Operate/Tasklist data.

Migration Considerations:

If you’re upgrading from 8.7 to 8.8:

  • The Importer is only needed temporarily to consume remaining 8.7 records from legacy Zeebe indices
  • Once all 8.7 data is imported, you can disable the Importer
  • The ES/OS exporters now export only a subset of record types by default (DEPLOYMENT, PROCESS, PROCESS_INSTANCE, VARIABLE, USER_TASK, INCIDENT, JOB)

For New 8.8+ Installations:

  • No Importer needed at all - the Camunda Exporter writes directly to Operate/Tasklist indices
  • You must use a single shared ES/OS cluster for all components (multiple isolated clusters are no longer supported)

So to directly answer your question: Yes, Zeebe now streams directly to Operate, which means Operate no longer needs to query and import from separate Zeebe indices for new data.

Would you like me to help you understand the specific configuration changes needed for your setup, or do you have questions about migrating from your current 8.7 configuration?

References: