Microservices Circuit-Breaker Pattern Implementation: Istio vs. Hystrix | Hacker News Reader

Microservices Circuit-Breaker Pattern Implementation: Istio vs. Hystrix

exoscale.com

86 pointsnfrankel19 comments

Hystrix is in maintenance mode - README says Netflix is moving towards e.g. https://github.com/resilience4j/resilience4j

> Netflix Hystrix is now officially in maintenance mode, with the following expectations to the greater community: Netflix will no longer actively review issues, merge pull-requests, and release new versions of Hystrix.

It’s mentioned in the article:

> Netflix has recently announced it has stopped development of the Hystrix library in favor of the less well-known Resilience4J project.

> Even if the client code might be a bit different, the approach between Hystrix and Resilience4J is similar.

These microservice design patterns bring back horrible memories of the design overarchitecture that was in vogue back in the J2EE years, where you basically need patterns to deal with the issues your new patterns bring up.

Of course in practice most reasonable people choose to deal with these things by dealing with complex failures in the most simple way, but the proliferation of these "solutions" IMO bring in more trouble than it could be possibly worth.

Also worth reading: https://vertx.io/docs/vertx-circuit-breaker/java/

And this: Pattern: Service Mesh - http://philcalcado.com/2017/08/03/pattern_service_mesh.html

That cookie notification is absolutely atrocious - is there really any need for a full screen popover??

Usually better to use asynchronous communication between services.

With a good old fashioned message broker for guaranteed eventual delivery, decoupling, discovery and elasticity (if you can fit it in to your setup)

I can't tell if this is a joke or not.

I'm not joking. Just to be clear, I'm taking asynchronous to mean a call that is not transactional. Where you are ok with eventual consistency. If something is transactional then keep it in one service. Make it an action not a transaction and save yourself a lot of grief.

So if something is asynchronous then why not back it up with a message broker and get the properties I listed (in earnest).

I'd be interested to know if anyone's had success with distributed transactions in a microservice architecture. I've gotten into a world of pain every time I've tried.

We did have some success at one of my previous companies (wiggle.co.uk), but that was mainly due to pretty tight integration between MSMQ, MSDTC and MS Sql Server. Even then, we tried to avoid distributed transactions as much as possible.

On its own this is a worthless hand wavy statement.

How do you accomplish this when you have a synchronous client (ie a user waiting in his web browser)? Should the browser poll for the result? How could you possibly implement something like this in the real world?

If it is synchronous or transactional then don't use microservices. Plenty of things can be done asynchronously provided the use case is ok with eventual consistency or within a non urgent time limit.

BTW now that the browser supports websockets (a bidirectional protocol the rest of the world had since the 70s!), you don't have to poll.

Your comment seems to be a slightly clumsy way of saying "I don't know much about this area of software design, please tell me more", so I'll respond to it in that spirit.

Firstly my terse hand wavey statement was at least clear on the fact that it was talking about communication between services. It said nothing about web browsers or any other human interface.

But since this relatively narrow context is clearly the one which you can most relate to, I'll stay within it. If you think about it, the HTTP request/response cycle is actually an illusion of synchronous communication on top of a transport mechanism which is fundamentally asynchronous. It's an abstraction layer which shields you from the complexities of asynchronous communication. Broadly speaking this is known as the [request/response](https://en.wikipedia.org/wiki/Request%E2%80%93response) pattern.

HTTP by no means the only implementation of the request/response pattern. Even in asychronous messaging systems it's not unusual to see it implemented - https://www.enterpriseintegrationpatterns.com/patterns/messa....

But request/reply is just one of the patterns which exists in the realm of asynchronous communication. There are other, much more common patterns such as send/receive and publish/subscribe. Exploring these patterns opens up whole new approaches to software design.

I wholeheartedly recommend the book [Enterprise Integration Patterns](https://www.enterpriseintegrationpatterns.com/) by Gregor Hohpe and Bobby Woolf. It's quite old now but is still the best and most complete introduction to asynchronous design.

An alternative perspective is from the viewpoint of [Command Query Separation](https://en.wikipedia.org/wiki/Command%E2%80%93query_separati...), and it's relative at the architural level, [Command Query Responsibility Segregation](https://martinfowler.com/bliki/CQRS.html).

It is almost never better to use async communication between services. The failure modes are strictly more, and finding and fixing problems is substantially harder.

https://programmingisterrible.com/post/162346490883/how-do-y...

Part of that article seems to be saying: I don't see any point in having suspension in vehicles. When I overload my truck it handles like shit.

Are you basing that on personal experience, or just on that blog post?

The blog post elegantly reflects my own experiences.

I found the blog post to be awful, the rationalization of a man stuck in a particular mindset who failed to adapt to a new paradigm. The line "You have publish-subscribe, but you really want request-response." says it all. He simply cannot imagine a software design which doesn't depend on synchronous request/response.

Reply on news.ycombinator.com