We built website

(Check it out: SleepySquids.com)

Websites are hard. We've collectively tried every type of hosting solution out there and they all have some drawback that makes them less than ideal. Usually that drawback is price, but in some cases it's technical limitation. Recently we spent a few days struggling to re-examine our hosting situation and find a way to support a devlog. The following is a summary of what we've learned during this round of searching, starting with a re-assessment of our needs.

What do we need?

Not every team has the same needs or wants from any given piece of tech, so we should first talk about what we were looking for. For developers who just need to make quick landing pages for small projects, or simple portfolio pages Carrd.io is a great choice. We still use them for some single pages and plan to continue for the foreseeable future. They have a great free option and a good balance of features and ease of use for paid customers. Our recent want for a devlog, though, meant we needed to find some other solution that would allow for easier management and updating of a multi-page site.

We're big fans of self-hosting and self-managing as many services as we can handle. If cost and time were no barriers we would have our own hardware co-located in a datacenter and manage our own stack of resources. The costs are not just monetary when talking about self managed solutions, though, as the time commitment of setting up and maintaining servers is significant. We've found that completely self managed systems are also the most prone to atrophy and tech debt over time as well. That means we're usually looking for solutions that allow us as much configuration control as possible, but don't require more time than we can spare on maintenance.

Of course managed services aren't without their own downsides. Services like wordpress or even the free tier of Carrd might require using a subdomain or leaving a "powered by" type of message in the footer that advertises their service. Seeing those footers has always left a very stale Web 1.0 taste in our mouths so it's something we wanted to avoid. We also have a few top level domains that we would like to continue using and prefer to manage our own DNS records.

It's also important to mention that we have zero collective skill at when it comes to web design. There has been a recurring conflict between our desire to host our own content and our inability to serve that content to the end user for some time now. Because of this any solution with a WYSIWYG editor is preferred.

So what we're looking for is something we can manage ourselves, easily expand into more complex infrastructure, and that supports a WYSIWYG editor.

What we tried

AWS offers nearly every service you can think of and we've tried a good number of them, so it was a good place to start. EC2 is the first thing we're often suggested, but in our experience they're usually not worth the expense. Any VPS solution will by design be over provisioned as well, meaning we'd be paying for capabilities we aren't using. The added maintenance of running a VPS combined with the higher cost ruled out EC2 for us. Lightsail is a similar service but costs a bit less and has a generous amount of bandwith allocation instead of a varying cost structure like EC2. These suffer from the other downsides of VPS hosting, though, and were equally ill-fitted for our current needs. 

One service we have used in the past with some success was to host static sites in S3 buckets, and this is what we expected we would end up doing. We found out shortly after setting it up, though, that getting SSL to work required the AWS CDN called CloudFlare. Cloudflare is another variable cost service from AWS that might have made S3 buckets technically feasible, but their payment structure was a big red flag to us. The risk that some malicious traffic could run up an AWS bill was too much for us, especially as new developers with nearly zero income.

The current solution

When S3 fell through as an option we were disheartened, but not in the worst position. By this point we had found an editor we really liked called Publii that basically lets you write WYSIWYG articles locally and push them to your static site host. It's like having a local Confluence site that you can push online with a button press, perfect for our time and skill constraints.

This left us looking for a static web host that supports HTTPS, is inexpensive, and is compatible with Publii. After a bit of searching we found DigitalOcean's App Platform, a service that takes github projects and automatically containerizes and deploys them in a managed environment. While it is meant for more complicated deployments it happens to be perfect for deploying a simple static site like this one. It also offers a free CDN, DDoS mitigation, and supports custom domains. The best part is since we don't need any CPU time or dynamic elements we can easily fit within the limits of the $0 starter plan they offer. If we decide we want some dynamic elements at a later date DigitalOcean has it's own Lambda equivalent called "Functions" that we can use - with it's own free tier of service, too.

Conclusion

While Carrd is certainly an awesome service we'll continue to use and recommend, the tech stack we've ended up with fits our (admittedly particular) needs perfectly for zero additional cost. We couldn't be happier. Years of struggling as hobbyists to host our own apps and content have paid off and allowed us to find a solution that serves our needs well. We hope our story helps you find a similar solution to your own needs; If you have any questions or comments feel free to comment, send us a tweet @Boppy, or join our discord and let us know.