微服务架构监控

Question

My company is in the process of migrating from a monolithic approach to a microservices architecture. I am wondering what tools are available to manage start up, shutdown, load balancing, and overall health monitoring.

The solution has to be windows based, but have the option to run in a linux environment as well.

At this time deploying locally is fine, we are not considering cloud at this time.

Thank you.

Answer 1

So, your requirements boil down to the following:

1) Environment/Infrastructure agnostic architecture that can run anywhere (local, VPS or cloud).

2) Service life-cycle management (start, stop, monitor,scale).

Agree with Heiko's answer that you should look for solutions built on Docker/Kubernetes. Such solutions need to extend the available primitives to provide the kind of solution you are looking for.

Hasura ( full-disclosure: I work here ) is one such PaaS that allows you to deploy your services on any infrastructure in an API Gateway pattern based architecture. Monitoring, autoscaling, and back-up & restore like features are available as add-ons. The platform is free to use on your own infrastructure. Check out a HackerNews discussion on Hasura .

Answer 2

I think your question includes a few concerns that may or may not be separate

1. running the microservices with lifecylce operations
2. health checks - is my service alive
3. telemetry

There are several solutions available where 2 and 3 build on 1 :

1. Orchestration systems like Kubernetes or the OpenShift distribution of it, take your µ-service, but it in an Docker image and then schedule it to run inside a cluster. There are other solutions like Docker-swarm that do similar things
2. Kubernetes and OpenShift can regularly "ping" your service. Be it via http or by invocation of a command. If this shows the service is unhealthy, then it is killed and a new instance deployed by the system. Similar to this liveness probe, there is also a readiness probe that can decide from what point in time on your service is ready to receive requests
3. There is a ton of metrics system out there that can retrieve telemetry data from the systems and the services. OpenShift for example comes with Hawkular Metrics out of the box. Data gathered can be graphed or alerted upon.