Soam Vasani on Fission

Using Fission's Prometheus Metrics

Tue, 16 Oct 2018 01:01:00 -0700

You can’t improve what you can’t measure. Visibility into metrics is a foundational requirement for much of software operations.

Serverless metrics collection can be tricky. For one, there is no “uptime” to measure. Secondly, with a “pull model” like Prometheus, there is no long-lived service to scrape metrics from. By the time Prometheus scrapes your function’s running instance, it may already be terminated, since functions are short-lived. Though Prometheus has a push gateway, it doesn’t maintain history, which makes it hard to track metrics for multiple concurrent function instances.

Automated Canary Deployments in Fission

Tue, 16 Oct 2018 01:02:00 +0000

Canary Deployments are a time-tested deployment strategy to reduce risk. The fundamental idea is that deploying software into a production cluster is different from releasing it to its users. With canary deployments, you deploy both old and new versions into a production environment, but send only a small percentage of traffic to the newer version. That way, if the new version fails, only a few users are affected rather than the application’s entire user base. If the newer version works well, the traffic proportion being sent to it is increased incrementally until the new version has been rolled out to all live traffic.

FaaS Function Composition with Fission Workflows and NATS

Thu, 22 Mar 2018 01:46:30 -0700

NATS is a lightweight, open source, high-performance, messaging system for cloud native applications, IOT messaging, and microservices architectures. The NATS messaging system implements a scalable publish-subscribe (or pub/sub) distribution model. There are a number of open source technologies like Kafka, and several cloud technologies such as Google Cloud Pub/Sub and Azure Service Bus that adopt this model as well.

In light of NATS recently being accepted into The Cloud Native Computing Foundation last week, we thought it’d be cool to share how Fission functions and workflows utilize this sleek messaging system on top of Kubernetes.