eCal vs Zenoh

[cvilas]

Hi all. Both ecal and zenoh are Eclipse-supported projects, and seem to provide similar functionality.

1. I just wanted to know what the outstanding features of eCal are compared to Zenoh. I write applications in C++ and Python, and don't care (at least yet) for Rust. Just want to understand where one might be better suited than the other.
2. Is there a development roadmap for ecal somewhere?
3. How do the developers see eCal evolving over time. What kind of use cases are targeted?

[rex-schilasky]

Hi @cvilas,

I intended to provide a detailed response but decided to see how ChatGPT would summarize it. Indeed, ChatGPT created an almost perfect summary! There’s not much to add or correct, in my opinion.

Welcome to eCAL! I hope we can support you and your projects in the future. In addition to its existing features and use cases, eCAL offers a robust set of ecosystem tools for recording, replaying, and monitoring your data flows. Its distributed recording design, in particular, is outstanding in the market.

Here the summary:

Eclipse Zenoh and Eclipse eCAL are both open-source middleware solutions within the Eclipse Foundation ecosystem, aimed at enabling efficient communication. However, they have different designs, architectures, and intended use cases. Here's a detailed comparison:

---

1. Purpose and Focus

• Zenoh:

  • Focuses on unifying data in motion, data at rest, and computations.
  • Bridges the gap between traditional publish-subscribe messaging systems, distributed databases, and high-performance computing.
  • Designed to scale across networks, from low-latency local environments to globally distributed systems.

• eCAL (Enhanced Communication Abstraction Layer):

  • Primarily designed for high-performance inter-process communication (IPC) in real-time systems.
  • Targets embedded systems and automotive domains with a strong emphasis on low-latency communication within a single machine or local network.
  • Well-suited for autonomous systems and robotics.

---

2. Architecture

• Zenoh:

  • Uses a peer-to-peer architecture with support for hierarchical storage and routing.
  • Supports data-centric pub/sub models, queries, and computations.
  • Can work seamlessly in constrained environments and over unreliable networks.

• eCAL:

  • Primarily relies on a centralized discovery mechanism for managing topics and services.
  • Optimized for shared memory and UDP multicast in local environments.
  • Focuses on deterministic data delivery with low jitter, ideal for real-time applications.

---

3. Features

Communication Paradigms:

• Zenoh:

  • Publish-Subscribe.
  • Distributed Queries.
  • Storage and retrieval of data at rest.

• eCAL:

  • Publish-Subscribe.
  • Remote Procedure Calls (RPCs) for service-oriented communication.

Scalability:

• Zenoh:

  • Scales across WAN and edge environments, with routing and bridging capabilities.

• eCAL:

  • Designed for LAN and single-host systems. It doesn't natively support global scale but excels in local networks.

Performance:

• Zenoh:

  • Efficient over diverse environments but may introduce higher latency compared to eCAL due to its general-purpose design and global scalability.

• eCAL:

  • Offers extremely low-latency communication and high throughput, making it ideal for performance-critical local applications.

Resource Constraints:

• Zenoh:

  • Optimized for resource-constrained environments, including IoT devices.

• eCAL:

  • Optimized for powerful embedded systems or desktop-class environments.

---

4. Use Cases

• Zenoh:

  • IoT and Edge Computing.
  • Large-scale distributed systems.
  • Environments requiring a mix of pub/sub and data persistence.

• eCAL:

  • Autonomous driving systems.
  • Robotics.
  • Real-time control systems.
  • Applications with strict latency and determinism requirements.

---

5. Interoperability

• Zenoh:

  • Flexible integration with various protocols and storage systems.
  • Extensible with plugins for different network topologies.

• eCAL:

  • Strongly integrated with Protobuf for data serialization.
  • Compatible with other frameworks like ROS2 through bridges but less flexible than Zenoh in heterogeneous networks.

---

6. Community and Maturity

• Zenoh:

  • A relatively younger project (since 2019) with growing adoption in IoT and distributed systems.

• eCAL:

  • More mature, with extensive use in the automotive industry and other performance-critical domains.
  • Backed by Rex Schilasky and Continental Automotive, with a strong focus on reliability in industrial applications.

---

Summary Table

Aspect	Zenoh	eCAL
Primary Use Case	IoT, Edge, and Distributed Apps	Real-time Local Systems
Architecture	Peer-to-Peer	Centralized Discovery
Latency	Moderate to Low	Ultra-Low
Scalability	WAN and LAN	LAN (Local Area Network)
Storage Support	Yes (Data at Rest)	No
Environment	Resource-Constrained, Global	High-Performance, Local

---

Choosing Between Them

• Use Eclipse Zenoh if you need:

  • A system that bridges messaging, storage, and computations across WAN.
  • Scalability to cloud and edge environments.
  • Flexibility in diverse network topologies.

• Use Eclipse eCAL if you need:

  • Deterministic, low-latency, high-throughput communication.
  • A middleware tailored for real-time applications in local or embedded systems.
  • Focused support for automotive or robotics industries.

[cvilas]

This is brilliant. Thank you.

[cvilas]

By the way, chatGPT gave me a similar answer. But given it's lack of accuracy it's good to have confirmation from someone from the project.

[rex-schilasky]

Yes for sure .. I was suspicious at first too, but the answer fit very well.