FluxNinja: intelligent load management for cloud-native apps (case study PDF)

FluxNinja's open-source project Aperture automates reliability processes for microservices, such as prioritized load shedding, intelligent autoscaling and distributed rate limiting.

This case study shows how CodiLime engineers classified and labeled each request with OpenTelemetry Collector pipelines and OPA Rego classifiers at no noticeable CPU cost, and covers the three ways Aperture runs: Envoy filter, SDK and sidecar.

The main topics covered in this report:

Optimized OpenTelemetry pipelines processing logs, metrics and traces
Request classifiers on default metadata using the OPA Rego language
Aperture as Envoy filter, multi-language SDK and VM sidecar
UI design and development for FluxNinja ARC dashboards
Team scaled to over 15 members in 3 months

FluxNinja: intelligent load management for cloud-native apps (case study PDF) — document render

Why download this report?

What is Aperture?
FluxNinja's open-source platform for load management in cloud-native apps. It automates prioritized load shedding, intelligent autoscaling and rate limiting through policies expressed as a signal processing graph.
How does a policy get the data it needs?
Requests are classified, labeled and processed through a configurable list of OpenTelemetry receivers, processors and exporters. The pipeline digests a great amount of traffic quickly and feeds each policy the data it needs to act.
What did CodiLime deliver?
Optimized OpenTelemetry pipelines for logs, metrics and traces, classifiers in the OPA Rego language, and Aperture as an Envoy filter, as an SDK in Go, Java, JavaScript/TypeScript and Python, and as a sidecar. The team also designed and built the UI of FluxNinja ARC.
What do Aperture's users get?
Aperture deploys on almost any infrastructure, and users pick predefined policy blueprints or create custom policies based on their own metrics.