Introduction:
Business process management (BPM) engines such as Camunda and Flowable are core building blocks for automating business processes in companies. Banks and financial companies in Switzerland in particular place high demands on such platforms – from reliable on-premise operation and scalability to compliance in regulated environments. A lot has happened since the original comparison of Camunda and Flowable: Camunda has introduced Camunda 8, a completely new, cloud-native BPM platform based on the Zeebe engine, while Camunda 7 (the classic Java-based engine) is slowly being phased out. This updated comparison looks at Camunda 7, Camunda 8 and Flowable in terms of CMMN support, architecture, scalability/performance, APIs & integration, licensing/enterprise features and on-premise suitability – with a particular focus on the needs of banks and insurance companies in Switzerland.
CMMN support: clarify Camunda vs. Flowable
One major difference in advance: CMMN (Case Management Model and Notation) is hardly pursued by Camunda anymore. Although the CMMN functionality has been implemented in Camunda 7, it is now only in maintenance mode – there will no longer be any new features camunda.com . In Camunda 8 is CMMN is no longer supported capitalone.com. Flowable, on the other hand, offers full CMMN support and continues to invest in this standard process forum.flowable.org
. For use cases such as complex case processing (e.g. credit cases, compliance checks) that benefit from CMMN, Flowable therefore has an advantage from the outset. Camunda itself recommends implementing such scenarios with BPMN (sometimes with pattern workarounds), but this can be cumbersome depending on the use case. Financial companies that rely on case management should pay particular attention to this difference.
Architecture: Camunda 7 (Java engine) vs. Camunda 8 (cloud-native Zeebe)
The architectures of Camunda 7 and Camunda 8 are fundamentally different. Camunda 7 is a monolithic Java-based process engine, originally a fork of Activiti. It typically runs within a Java application or on an application server and uses a relational database to store process instances, tasks and historical data. You could run Camunda 7 embedded in your own Java code, as a shared engine on a server or as a separate service with a REST API, depending on your needs blog.viadee.de . This flexibility allowed banks, for example, to integrate the engine directly into existing core banking applications or to host it as an independent BPM service in their own infrastructure.
Camunda 8 takes a completely different approach: it is based on the newly developed Zeebe engine, which relies on distributed microservice architecture rst.software. Instead of a single Java engine with a shared database, Camunda 8 consists of multiple components (Zeebe Broker/Gateway, Operate, Tasklist, etc.), which are typically orchestrated in a Kubernetes environment blog.viadee.de. The engine is designed for this, horizontally linear scaling and does not require a relational database – the classic DB bottleneck of Camunda 7 is eliminated blog.viadee.de. This means that Camunda 8 can handle considerably more process instances in parallel without being tied to a central DB transaction. However, there is the Camunda 8 engine only as a remote serviceEmbedding in own applications as with Camunda 7 is not intended forum.camunda.io. Process execution is always decoupled in remote mode – Camunda 8 only orchestrates the order of the tasks, the execution of the task steps takes place externally (similar to the External Task Pattern of Camunda 7)blog.viadee.de.
The Flowable engine is a lean Java process engine core that also runs in a classic multi-tier architecture with a relational database. It can be operated embedded in a Java application or as a separate service in the server/cluster and integrates perfectly into Spring-based environments github.com . With regard to the Architecture This means that Flowable can be deployed as flexibly as Camunda 7 (e.g. in a bank’s on-premise server landscape), while Camunda 8 requires a significantly more complex set-up with distributed components. For conservative IT environments in banks, which often still rely on proven architectural principles, the steeper learning curve of Camunda 8 could play a role.
Scalability and performance
The different architectures result in major differences in scalability. Camunda 7 traditionally scales vertically and horizontally to a limited extent – i.e. you can run the engine on more powerful hardware or in a cluster with a shared database. However, the latter reaches its limits at some point, as the relational database is the limiting factor for the number of simultaneously running processes blog.viadee.de . For many typical applications (including most internal bank processes) Camunda 7 is performant enough, but with extremely high load peaks or mass data, performance can drop if the DB becomes a bottleneck.
Camunda 8 was explicitly designed for high throughput rates and linear scaling. The Zeebe engine uses an event streaming and partitioning model that allows additional broker nodes to be added and processing to be distributed across multiple instances – without a central transaction database. In practice, Camunda 8 can therefore process significantly more process instances per second than Camunda 7 blog.viadee.de . This scalability can be a decisive factor for financial companies with digitization projects that have to handle very large volumes (e.g. in payment transactions or real-time processing). However, it should be noted that Camunda 8 is based on possible consistency is set: Changes are not immediately visible in all components, which must be taken into account in strongly interlinked synchronous processes.
Flowable‘s performance is comparable to Camunda 7, as it is also based on the established transaction-based model with relational DB. For most business workflows – such as credit decision processes or account openings – Flowable is sufficiently scalable. In load tests with thousands of simultaneous instances, performance depends primarily on the database and the tuning of the environment, as is the case with Camunda 7. Flowable does not offer a special, newly developed “high-throughput” mode like Zeebe, but the engine is designed to be “lightweight” and can run efficiently on modern hardware or in container orchestrations. In short: Camunda 8 has the edge when it comes to massive scaling, while Camunda 7 and Flowable are performant and proven for “normal” process loads. In regulated industries, where stability is often more important than maximum throughput, the advantage of Camunda 8 is put into perspective – unless the deployment scenario really requires the handling of very large volumes of events in a short period of time.
API and integration options
The decisive factor for developers and architects is how the BPM engine can be integrated into the existing system landscape. Camunda 7 traditionally scores points here with direct Java APIs and a REST API. In embedded use, Java delegates can be executed directly in the same JVM process, which allows close coupling and even joint transactions with the application. Alternatively, Camunda 7 offers a full REST API that external applications can use to start processes, query or complete tasks. This flexibility made Camunda 7 attractive for many banks, as it was possible, for example, to integrate Java modules with Camunda in the core banking system and at the same time access the engine in other areas (such as an online banking platform) via REST.
In terms of integration, Camunda 8 relies on a dedicated gateway with gRPC API. Instead of talking directly to the engine (which is not possible due to the decoupling), all communication runs via the Zeebe gateway and its gRPC interface blog.viadee.de . This channel can be used to start processes, pick up tasks and complete them. As gRPC is less common in the Java ecosystem, Camunda provides official Client libraries for Java, JavaScript, etc. that facilitate communication blog.viadee.de. In addition, the interfaces in Camunda 8 are distributed across several services – e.g. there is a GraphQL API for the tasklist and a REST API for the Operate tool (process monitoring) blog.viadee.de. For integrations, this means a little more effort and new technologies (e.g. gRPC, GraphQL) compared to the familiar REST/Java world. In a modern microservice architecture, however, these patterns are easy to connect. A side effect: Since Camunda 8 always works remotely, there are No Java delegates or direct execution listeners more in the engine process – such logic must be outsourced to external workers forum.camunda.io.
Flowable offers a very developer-friendly model in terms of integration. Like Camunda 7, Flowable provides Java APIs and can be seamlessly integrated into Spring applications github.com . At the same time, there is also a comprehensive REST APIso that Flowable can either be addressed via HTTP or used as an embedded library – depending on requirements. This duality is ideal for many banks: for example, a core banking system can use Flowable as an integrated workflow orchestrator, while other satellite systems communicate with the workflow service via REST. Message queue or event integrations can also be realized, similar to Camunda 7 (e.g. External Task Pattern can be implemented analogously in Flowable). Overall, Flowable requires No new protocols or frameworks in integration; developers can rely on familiar technologies. This is a plus point for existing Java stacks in financial companies, as the learning curve is lower. Camunda 8, on the other hand, comes with more modern interface technologies, which can be advantageous in future-oriented architectures (keyword cloud-native banking) – provided the know-how for gRPC & Co. is available.
Licensing and enterprise functions
License models and available enterprise features also play a major role in the choice of BPM engine – especially in large financial institutions that need support and long-term planning security. This is where Camunda 7, 8 and Flowable differ considerably in some cases.
Camunda 7 was offered in two editions: the Community Edition (CE) is open source (Apache license) and can be used free of charge, while the Enterprise Edition includes a paid subscription with support and additional features (e.g. extended cockpit functions, LDAP integration, Camunda Optimize for reporting, etc.). Many companies – including banks – initially evaluated Camunda 7 in the free version and then upgraded to the Enterprise version for productive use. Important to know: Camunda has announced that it will only actively support Camunda 7 until around 2027 (community updates only until October 2025) pretius.com . After that, full support ends (extended support with bug fixes only until 2027). New customers are clearly steered in the direction of Camunda 8as Camunda 7 is still being maintained but is no longer undergoing any further development.
Camunda 8 also breaks with the tradition of open source when it comes to the license model. The new platform no longer has a free community edition, but always requires a commercial license for productive operation pretius.com . Although the source code of the core engine Zeebe is publicly available, the use of the entire platform (incl. Operate, Tasklist, etc.) is only permitted in production with a valid contract. Camunda 8 is available both as SaaS (Camunda Cloud) as well as offered as self-managed variant for own data centers, each in staggered enterprise packages. Free use is only possible for development/testing to a limited extent pretius.com. For banks, this means If you want to rely on Camunda 8, you have to plan a budget for license & support. In return, however, you get access to all new components, security updates and Camunda support. The Enterprise functions in Camunda 8 include Operate (process monitoring similar to Cockpit), Tasklist (task UI), Optimize (BI/reporting) and Identity (rights management) – many of these were only available as add-ons or in the Enterprise version in Camunda 7. In Camunda 8, they are an integral part of the offering (but not freely selectable as modules in the community, as these do not exist).
Flowable follows a dual model similar to that of Camunda 7: There is an open source engine (Flowable Community) under Apache 2.0 license github.comwhich can be used and customized free of charge. In addition, the company Flowable offers Enterprise products (e.g. Flowable Work for case management with UI, Flowable Engage for chatbot integration etc.) that are commercially licensed are forum.flowable.org. These enterprise tools are based on the open source core, but offer convenient web interfaces, integrations (e.g. Elasticsearch-based search) and additional support for operations. In principle, a financial company can therefore start with the free Flowable engine and build its own solutions around it – or use the Enterprise package in order to obtain ready-to-use solutions (incl. modeling UI, admin UI, etc.) more quickly. The costs for Flowable-Enterprise are individually negotiable; they tend to be regarded as more flexible and often cheaper compared to large BPM suites. Nevertheless, you should bear in mind that the open source version of Flowable does not have limited engine functions (everything, including CMMN, is included), but for professional use in a bank, the enterprise extensions for monitoring, scaling (e.g. multi-node tools) and support are often very valuable. This depends on the bank’s requirements and resources: If you have a strong in-house development team, the open source version may be sufficient; if you need fast results and manufacturer SLA, the enterprise version makes sense.
To summarize: Camunda 8 is fully committed to a commercial subscription model, while Flowable is sticking to the open source route and has optional enterprise offerings. For a bank with strict compliance requirements, open source can be attractive (keyword: independence, no vendor lock-in), but at the same time you have to weigh up whether you can make up for the lack of official support internally. Camunda is very well established as a commercial provider in the BPM world, which also means that expertise could be easier to find on the market. Flowable is somewhat less well-known, but also has an active community and is supported by former Activiti developers – the code base is tried and tested.
On-premise operation and relevance for regulated industries
As a highly regulated industry, aspects such as data sovereignty, security and compliance are crucial for Swiss banks and insurance companies. Many institutions prefer (or require) on-premise operation of their core systems in order to have full control over data and processes – cloud solutions are often only permitted hesitantly or in isolated areas, especially when customer data or transaction-relevant data is involved. Against this backdrop, the operating mode of the BPM engine is a critical factor.
Camunda 7 is predestined for on-premise from the ground up. You can run the engine in your own infrastructure – be it on dedicated servers, in virtual machines or containerized in your own data center. All components (database, web apps such as Cockpit/Tasklist etc.) run under the full control of the bank. This fits in well with the usual IT requirements in banks, where software is ideally operated within the bank’s own security zones. Camunda 7 can be used in highly secure environments (with network segmentation, without Internet access) without any problems and meets the requirements for auditability and traceability set by FINMA, for example.
Camunda 8 was primarily developed with a cloud strategy in mind. The standard offering is a SaaS service, hosted in the provider’s cloud (Camunda Cloud). This model is often problematic for banks, as process data would be stored and processed outside of their own infrastructure. Although Camunda Cloud guarantees operation in European data centers and offers encryption and ISO certifications, many financial companies have policies that prohibit the outsourcing of such workflows to a public cloud. The good news: Camunda 8 is also available as a self-managed version that can be operated on your own servers (or in a private cloud) blog.viadee.de.
In practice, however, this means increased effort: a bank would have to provide a Kubernetes cluster environment, install, configure and operate the various Camunda 8 components. Not every IT operation is already set up to manage a distributed microservice platform internally. In addition, Camunda 8 Self-Managed currently still has a few limitations (as of today, some SaaS features are only available on-prem to a limited extent). For Regulated industries However, the self-managed route is usually the only viable option if you want to use Camunda 8 – and it is generally feasible (some large banks and insurers are already evaluating or using Camunda 8 on-premises, often in container platforms such as OpenShift). Important: The bank is still dependent on the manufacturer, especially when it comes to updates and support, as the platform cannot be operated without a license.
Flowable is an on-premise / self-hosted solution by origin. Designed as an open source engine, Flowable runs in any environment – whether classically on an application server, integrated into an existing application or in Docker/Kubernetes within the bank. There is no obligation to use any cloud service. This gives financial companies maximum sovereignty: all process data remains in-house and adjustments can be made to the code if necessary. Flowable, Inc. itself does not (as far as is known) offer its own public cloud BPM SaaS – the business model is rather to sell licenses for the enterprise tools or to support projects. Flowable therefore fits very well into strictly regulated IT environments. For example, Zürcher Kantonalbank (ZKB), the largest cantonal bank in Switzerland, has chosen Flowable as its process automation solution to significantly accelerate internal processes flowable.com . In this case, Flowable could be implemented on-premise, which meets the high data protection requirements. Especially when a bank CMMN case management (e.g. in loan processing or in the compliance environment), Flowable is a very attractive option due to its complete support and on-premise capability.
To summarize, banks and insurers need to evaluate for themselves: Camunda 7 can be operated on-prem without any problems, but will be replaced in the future. Camunda 8 offers modern cloud technology and enormous scaling, but requires a modern containerized operation on-prem and does not come without commercial contracts. Flowable can be used completely on-premises (open source), which allows maximum control – but here, too, you should consider the enterprise-level tools for productive use. In Switzerland, where regulatory requirements are strict, Flowable could therefore be the more pragmatic option in some cases, provided the functions are sufficient and there is no problem with the smaller manufacturer. Camunda 8, on the other hand, may be of interest to institutions that deliberately pursue a cloud-native approach (such as new digital platforms detached from the core banking system) or that need to map extremely high loads, where the architectural advantage comes into play.
Comparison table: Camunda 7 vs. Camunda 8 vs. Flowable
For a better overview, the following table summarizes the most important criteria and differences between Camunda 7, Camunda 8 and Flowable:
Criterion | Camunda 7 (Platform 7) | Camunda 8 (Platform 8) | Flowable (6 / V7 Engine) |
---|---|---|---|
BPMN 2.0 support | Complete (stable for years) | Complete (new Zeebe engine) | Complete (BPMN 2.0) |
CMMN support | Yes, but only maintenance mode (no further development) | No (not available in Platform 8) | Yes, fully and actively developed |
DMN support | Yes (Camunda Decision Engine 7) | Yes (new integrated decision engine) | Yes (embedded in engine) |
Architecture | Classic Java engine, monolithic or embedded, requires relational DB | Microservices architecture (Zeebe Broker/Gateway + Services), event sourcing, no relational DB | Classic Java engine, runs embedded or as a service; uses relational DB |
Scalability | Vertical / limited horizontal scalability (DB as bottleneck) | Highly scalable (horizontal scaling through distributed brokers, no central DB) | Similar to Camunda 7: limited horizontal (DB cluster), sufficient for most use cases |
Performance | Proven, good performance up to medium load; DB limitations for very high loads | Very high performance even with a massive number of instances (possibly consistent processing) | Good performance for normal load; light footprint, efficient engine |
Integration options | Java API (embedded), JUEL scripts, REST API, external tasks; direct transactions with app possible | gRPC API via Zeebe Gateway (clients for Java, JS etc.) REST/GraphQL for individual components; always asynchronous (external task principle) | Java API (embedded possible), REST API; seamless Spring integration synchronous & asynchronous possible (queues, listeners etc.) |
Enterprise features | Cockpit (limited in Community), complete Operations UI & Admin only in Enterprise; Optimize (Reporting), Consultants Support etc. | Includes all new tools (Operate, Tasklist, Optimize, Identity) out-of-the-box in Enterprise; continuous updates from the manufacturer | Flowable Work/Engage (licensed UIs for cases, chat, etc.) ; Open source engine contains all core functions, but fewer convenience tools |
Licensing | Open Source (Apache 2) for Community Edition; Enterprise Subscription for full functionality; Camunda 7 CE EOL October 2025 | Enterprise license model only (no free prod use) SaaS or self-managed with support contract | Open Source (Apache 2) for Engine ; Enterprise licenses required for additional products/support |
Operation / Deployment | On-premise (own servers/VMs, containers) or platform-as-a-service by the bank itself; simple installation (one Java process + DB) | Default as Camunda SaaS cloud; alternatively self-managed on-prem in own Kubernetes environment more complex installation (several services) | On-premise (server, VM, container) fully supported; no official public SaaS by manufacturer (Design Modeler available as a cloud service) |
Suitability for regulated environments | High – complete data sovereignty on-prem, proven stability, audit logging available | Limited – only self-managed option on-prem suitable (higher effort, license requirement); SaaS not permitted for many banks | High – complete control on-prem, open platform (source code can be viewed); compliance adjustments possible; already used by Swiss bank |
(Legend: BPMN = Business Process Model and Notation, CMMN = Case Management Model and Notation, DMN = Decision Model and Notation)
Conclusion: Decision criteria for financial companies
The choice between Camunda 7, Camunda 8 and Flowable depends heavily on a company’s individual requirements and strategies. In the financial sector – where reliability, compliance and data control are priorities – the following considerations should be factored into the decision:
Modernization vs. proven: Camunda 8 represents a modern, cloud-native approach with high scalability and forward-looking architecture. Companies that rely on microservices and cloud technologies in the long term (perhaps as part of a digitalization strategy or to cover very volatile loads) will find a powerful platform in Camunda 8. However, the maturity of a new engine must be considered – Zeebe is still relatively young (developed since around 2017) blog.viadee.de, while Camunda 7 and Flowable are based on engines that have been tried and tested for years. based. For conservative use cases, “proven stability” could be more important than cutting-edge technology.
Functional requirements (BPMN/DMN/CMMN): If case management (CMMN) is an essential part of the requirements (e.g. in complex case processing that requires flexibility), Flowable has a clear advantage due to its complete support. Camunda 8 would be completely ruled out in this respect, Camunda 7 only with workarounds in BPMN capitalone.com . For pure BPMN workflow orchestration with Decision Tables (DMN), all three solutions offer solid functions. Camunda 8 integrates DMN directly into process flows somewhat more elegantly than Camunda 7 (automatic triggering of decisions)
– However, this is not a game changer for most financial processes, as DMN can also be used in Camunda 7/Flowable without any problems.
Integration effort and technology stack: Banks with large Java-based application landscapes and many legacy systems may prefer the seamless Java/REST approach of Camunda 7 or Flowable. Getting started is quick here: developers know the APIs, there are no surprising new technologies. Camunda 8, on the other hand, may require new skills (gRPC, Kubernetes) – which may be an investment in modern skills in the medium term, but increases the hurdle in the short term. If a bank is already moving strongly in the direction of cloud native (container platform available, event-driven architecture established), Camunda 8 is of course easier to integrate. If not, Camunda 7 or Flowable could lead to the goal more quickly in the current state.
Operating and license costs: For many financial decision-makers, TCO and compliance costs are crucial. Camunda 7 Community was attractive because there were no license costs – this option is no longer available with Camunda 8. Camunda 8 therefore means license costs (and not too low, depending on the number of instances/cluster size). In return, however, you get support and further development from the market leader. In principle, Flowable offers the opportunity to save on license costs (use open source), but this can be offset by higher internal costs. A bank that wants to be more independent of vendors and may have its own development teams can take a cost-effective route with Flowable Community. However, if dedicated support and guarantees are required – which is often the case in critical banking processes – you should also consider the enterprise version of Flowable. Overall, the commercial Camunda 8 packages are probably more expensive than a Flowable Enterprise license, as Camunda, as the market leader, has positioned its prices accordingly. Here it is worth getting quotes and also looking at which features are included in the price.
Strategic roadmap and vendor dependency: Camunda 7 is on its way to the end of its life cycle (end-of-life). Anyone still using Camunda 7 today (e.g. because it currently meets their needs) needs a migration strategy for the future – either switching to Camunda 8 at a later date or switching to an alternative. Camunda 8 is being intensively promoted by Camunda and is likely to remain the flagship for the next few years. Flowable is also continuously developing its engine and has made a clear commitment to CMMN and classic BPM technologies
. Vendor dependency also plays a role here: Camunda as a company has a strong presence and community (many extensions, forum posts, conferences, etc.), Flowable is somewhat smaller, but specialized. Some banks tend to rely on the “market standard” (which would currently be Camunda), while others look specifically for niche solutions that fulfill their exact purpose.
Conclusion in a nutshell:
- Camunda 7 is a robust choice for today’s on-premise requirements, especially if you need a proven BPM solution quickly and CMMN is not an issue. However, it is technologically at the end of its innovation cycle – for long-term projects, the foreseeable replacement should be taken into account.
- Camunda 8 is ideal when very high scalability or a cloud-native orientation is required. Camunda 8 can set the right course for new digital initiatives (e.g. API-orchestrated processes, high volumes in real time). In strictly regulated environments, however, the additional costs for self-managed operation and license dependency must be taken into account.
- Flowable is ideal for flexible on-premise solutions with full BPMN/DMN/CMMN functionality. It is a cost-effective alternative that is already being used successfully by banks (e.g. ZKB in Zurich)flowable.com. Companies that want maximum control and CMMN functionality will find Flowable a suitable candidate. However, one should be realistic about the fact that additional development work may be required to replicate the convenience of a Camunda Enterprise solution.
Ultimately, there is no “one size fits all” solution. Each institution should evaluate the options based on its own criteria. A proof-of-concept phase with the favored engines often helps to see which one fits best into the existing landscape. It is also worth taking a look at current analyses and community comparisons – for example, in 2022 a tech blog by Capital One compared various open source BPM systems (Camunda, Flowable, etc.) in detail capitalone.com . It is important to keep an eye on the long-term consequences of the decision: Technology stack, maintainability, costs and regulatory compliance. With the information and criteria presented here, banks and financial companies in Switzerland are now better equipped to make an informed decision between Camunda 7, Camunda 8 and Flowable