Hacker News Reader

Comment by ossinnameonly | Hacker News Reader

In my estimation, now would be a rather poor time to adopt SF.

1. It's poorly positioned against K8s. The "it can do containers" people are really not helping their case; SF's value proposition is clearly not in running arbitrary containers. Unless you're writing a new .NET app and will go all in on their SDK and replicated state model, it just doesn't make sense.

1a. It has no community. It has no presence outside of MS doc pages, the repo they threw over the wall, and the one external user that I'm aware of (who is already in this thread).

2. SF was operationally not-fun. I'm being nice. Though, neither is Kubernetes, most days.

3. SF is losing to Kubernetes, even internally, and even despite pressure from high-up-ish people.

4. The rumors about SFs fate would not seem to lend themselves toward adopting it for new projects.

View on HN

I don't know the first thing about SF (the product), and you might be absolutely right. But I find it discomforting to base engineering decisions on the equivalent of Facebook likes and Web presences/Googling rather than merit. For all I know, the media presence of the space of cloud orchestration software is highly manipulated by huge PR budgets (for example by Docker) and by media heavyweights competing in this space.

I agree with the sentiment, but what’s sometimes overlooked is that the community around a technology is in many ways more important than the technology itself. Can I expect to find add-on libraries that help integrating with the rest of my technology stack? Will I find documentation, blog posts or StackOverflow answers that solve my specific edge case? How likely is it that I’ll be the very first person to encounter a critical bug, because no one else uses it at that scale?

Many of those signals are not easily faked, as Microsoft itself painfully learned when spending billions in an attempt to artificially prop up a developer community around Windows Phone. If, as the grandparent suggested, Service Fabric is losing out even internally at MS where they can apply management pressure in addition to marketing, that’s a very bad sign.

Since my company uses .NET and a lot of MS products, I was tasked with learning SF in early-mid 2016. I don’t remember specifically what I didn’t like but it was probably poor documentation and slow build/deploy. It didn’t feel production ready.

Shortly after, we adopted Docker which was nice because we could run some Node.js services and some .NET Core services.

Although Docker introduced its own problems, namely orchestration.

I just replied to someone else about this but the article is about the internals. That is the interesting part.

I don't care about the "k8s 4 life" thing that other people seem invested in. The space that SF is useful in n is pretty large. There's room for multiple options.

Reply on news.ycombinator.com