Reading Time: 3 minutes
DiffusionData announced today the release of Diffusion 6.10. Diffusion 6.10 is the latest version of Diffusion, a framework that focuses on developers. It aims to reduce costs and increase productivity by reducing development time and enhancing developer experience.
Riaz Mohamed, CTO of DiffusionData said: We spend lots of time talking to developers and architects on how we can improve the framework. What do they need and are we not providing? And what can we do differently? This release was largely based on feedback that we received. These enhancements are designed to address the architectural and technical issues that many of our clients face every day. “We are grateful for all the input from our clients and community, which allows us to strengthen our position as a leader in real-time streaming data.”
The following enhancements are included in Diffusion 6.10:
Support for Floating Points in Topic Views
Topic Views now supports floating-point numbers for all clauses. This is especially true for the ‘process transformation’, which now supports floating point numbers in calculations. The ability to specify scale (the number after the decimal place) and rounding policy is included.
Operator Constraints
Diffusion allows you to specify restrictions when updating a subject, whether directly or via an update stream. Checking the current value before updating the topic is one of these constraints. Up until now, the only way to constrain a value’s value was by comparing it with its current value in binary. You can now update multiple topics at once, while maintaining the order. This is a response to a customer request.
The 6.10 version now supports relational comparisons such as “greater than”, “less than”, etc. It is not necessary to match exactly the type. You can, for example, specify that an update will only be applied if the value of the current value falls below five. This would work with integral and floating point values. This would work with a string containing a numerical value.
In the past, it was possible for update constraints to be composed using the and methods so that an update would only occur if the lock was held or if the value equaled a specific value. You can now use and to create constraints in 6.10. This will allow you to specify that the update only occurs if the topic is empty, has no value, or if the value is below five.
Retrieve metrics using the API
Metrics give users valuable insight into the performance of their system and allow them to fine-tune their installation. Prior to 6.10, metrics were available via Prometheus and the Diffusion console. In 6.10, you can retrieve metrics using the API.
The metrics features has a new method metricsRequest that returns an object you can configure. You then use a fetch to get the metrics. It will by default return metrics for all cluster servers, but you can request metrics only for the currently connected server or specified servers. The user can filter metrics to only show metrics that are relevant. The API provides metrics access, allowing users to integrate metric monitoring in their own applications.
Update Stream Recover
Clustered architectures offer the best availability, but can be problematic when the shape changes (for example adding a server). This feature allows for seamless recovery of applications from cluster blips. They would have had to code their applications to be resilient against these issues. It is less code to maintain and write for the user.
A transient error will occur if a Diffusion Cluster repartitions while clients are updating. The update stream will be invalidated. It would be necessary to code the updating application to recover in such a case. In 6.10, it is possible to specify a retry strategy and a new RecoverableUpdateStream is created. This stream can reapply in-flight updates when a transient error occurs.
RESTful API to Manage Gateway Framework
Gateway Framework allows for easy integration with numerous external systems. Diffusion console was used to manage the previous gateway applications in real-time. With the addition of REST-API management, such applications can now be managed programmatically within a user’s application. The REST API authenticates the user using a Logon request and returns a token that can be used to make REST calls in order to manage the Gateway.
