CloverETL 4.9 is all about Server robustness. It brings a major architectural change under the hood of CloverETL Server, aimed at increased fault tolerance and stability in mission critical deployments.
With 4.9, we are splitting the Server's internal architecture into two units. Don't worry though—for users and admins the change is almost invisible, most of the new stuff happens under the hood. First, we have a "worker" runtime module that executes transformation graphs and jobflows. The rest of the functionality of the Server remains in the "core" module (Server Web Console, user management, scheduler, data services, etc.).
By isolating job execution into a separate process of the host operating system, we are making the Server resilient against potential issues in individual jobs and we allow users more convenient control over resources the worker gets (e.g. user-friendly memory settings).
This new architecture is backwards compatible with previous CloverETL versions and does not require additional modules or licenses.
Server's internal architecture has been split into two separate modules, each running in a separate OS process. First, a "worker" runtime, executes transformation graphs and jobflows and the second module, "a core" manages all other functions of the Server (Server Console, user management, scheduler, data services, etc.).
By isolating job execution into a separate process of the host operating system, we are making the Server resilient against potential issues in individual jobs and we allow users more convenient control over resources allocated to the worker (e.g. user-friendly memory settings).
This under-the-hood change is designed to have minimal impact on users. No changes to your projects and or package installation or licensing are required.
You will notice new "Worker" panel under Monitoring section and new configuration options under Configuration/Setup. See the updated installation instructions for more details.
This new architecture is backwards compatible with previous CloverETL versions and does not require additional modules and licenses.
For users who manage large numbers of schedules in the Server, we're introducing filtering and paging of the Scheduling module, allowing for quick access to individual scheduled jobs.
In the past, internal buffers in the Runtime used so called Java "direct memory". This brought us marginal performance improvements but proved to cause trouble in many different Java setups.
Direct memory (USE_DIRECT_MEMORY) is now disabled by default. However, you can switch it back on if you wish to. Refer to Engine Configuration documentation for details.
Please pay close attention to Compatibility Notes when upgrading to 4.9 from previous production versions or Milestone releases.
To help you install or upgrade to this version, we've prepared a simple checklist:
Before You Upgrade
|Compatibility/Upgrade notes, Features & Fixes
Features & Improvements
|August 14, 2018
|Download CloverETL 4.9 4.9.1
Server Robustness - Worker
Server Robustness - Worker (changes from Milestone 2)
|July 23, 2018
|Download CloverETL 4.9 4.9.0