genox.ch

16.12.2018
Filed under [object Object], [object Object]
9e3759bd-56f2-4477-bf17-8ffa68bd4784

zeit.co and now v2

I got to know "now" by zeit.co as a dead simple way to deploy docker containers with everything needed for small to medium projects - for a price that no other service was able to match. Including DNS and CDN handling. No container registries. No huge configs. No AWS potpourri of services. No insanely high container prices like GCloud. No "let's fill it to the brink and make it look like Office" Admin UI like Azure.

Install a CLI utility, add a simple json config file with 6 lines to a project, add a domain and off you go. Even integrates nicely with build pipelines. All there. I thought I finally found the thing I will stick with, leaving behind shared hosting servers and LAMP altogether. Oh the simplicity.

This was in May 2018. Fast forward to August 2018 and lo and behold, it even got better: Serverless Docker (dubbed "now v2 beta" ) on the same platform, all it requires you to do is change a few lines in the config file and redeploy. I thought it was not necessary (after all, things ran smooth :-P). But still, I liked it. Again, fast forward to November 2018: Zeit announces now v2 officially and I couldn't wait to find out what changed since the beta. Quite a lot, in fact: Docker is no longer mentioned and v1 (what I came to love so much) is marked as deprecated. You're kidding, right?

Apparently, it's all about serverless now. Functions, functions, functions. But that's not why I started using it. That's not the product they advertised just a few months earlier. Not what I signed up for. In a matter of 8 months, they pivoted to a point where I have to revisit all the decisions I've made prior to migrating apps. The product I chose for its advertised use case suddenly is completely different.

Meanwhile, Zeit's Spectrum channel is exploding with people mentioning their frustration and worries. The company's CEO responds to every critical comment with a mix of understanding and a subtle nudge towards "follow us, the sights will be brighter, the forests more lush than ever before". I respect that, can understand it to a certain degree. But.. that's not what I expect to hear from a company I trust with running all the products my team works on.

We make decisions based on project requirements and personal preferences, yes. But we rarely rewrite an entire project after it being released just because it's "fun". No benefits. Meanwhile, the V2 being launched cannot replace most of the use cases their Docker service could handle easily, or even just pushing a directory with static files - doesn't work. Each static file that is available via a URL needs to be manually entered in a config file. You can't be serious. I don't know man, tell me that I make a mistake, but this sounds utterly rushed and driven by external forces or "investors concerned with company turn over", not by "affection for technology". The change from "everthing is in a container" to "everything is a function" is a massive, massive change. A change that sets in motion so many other concerns.

Why on earth would a company kill (doesn't matter that there's a transition period, it's two different products) a service that so many people came to love and advocate professionally? Something that stands pretty much alone on the entire market? A market that is very big with a product that scales perfectly?

I don't know man. I just wish someone else will take over that position and offer a similarly rich UX for the entire deployment process like Zeit.co did with Now v1. I for one will not rewrite all my stuff to work with a proprietary lambda service. It doesn't add anything to my projects. The only thing it creates is needless iteration and a massive technology break.

The only perspective this makes sense from is that Now wasn't profitable due to large infrastructure and maintenance. The service really is quite affordable, 10 containers for $15 bucks a month - go find that elsewhere, and all the nice CLI goodness, hmm.. I wonder how they made that work. Knowing that many SV startups and companies don't necessarily look at profitability in the beginning, one thing I could assume is that some of their investors weren't really happy with their numbers and they had to rush out a service that integrates with (it all runs on AWS Lambda) existing solutions to lower self implicated maintenance and scaling issues. But that's completely speculative.

Anyways - Now v1 will be deprecated. If they do not decide to rebrand V1 to something else and keep it around, I will look for alternatives. Since I read about the release of v2 I was digging around and man is it hard to find something similar. The only thing close to Now is Sloppy.io - which sufferred a 48 hours downtime in spring.. Not what I'm looking for either. And I really don't want to mess around with deployments on a lower level. That's the only way deploying Node based Webapps is financially feasible for small/mid sizes projects maintained by small teams.

All in all, it feels like Zeit is throwing away a completely fine and working product and replaces it with something that's fancier on many levels but doesn't even match the use cases of the former. I'm not aware of many companies who could actually survive this. It just feels off. And even if they decide to offer docker deployment as its own product in the future, I'm not sure I can trust a company that pivots so much in a field where everybody only wants one thing: a stable, predictable way to run a container without the hassle of maintaining container infrastructure.

1