Recently I asked solo-founders what are their choices when it comes to hosting their projects. Here are the most common choices along with their rationales.
When it comes to hosting our projects we have several options:
- platforms
- cloud infrastructure
- serverless
Platforms
Heroku
Heroku has been a very popular option for hosting projects for several years now. It’s very easy to set up and has a wide range of plugins such as PostgreSQL, Redis or NewRelic. It also has a huge community raised on building products on Heroku.
However, it can get fairly expensive as you scale your app. Their resource tiers are also very arbitrary and sometimes force you into paying for way more than what you actually need. From my personal experience, reliability on Heroku can also be a pain.
Netlify / Vercel
Netlify and Vercel seem to have a pretty similar offering. For some reason, I see Netlify more often, but Vercel as the creators of Next.js framework are not to be dismissed. If you want an in-depth comparison of Vercel and Netlify, Lambros Petrou wrote a nice post summing up their capabilities.
Both options are a very very popular option for hosting static pages – and for a good reason. Easy to deploy, both feature built-in CI/CD along with commit previews. And of course, they employ a fast CDN so that your site load times are very small.
If you’re into serverless (which really makes sense as a solo-founder), they also have the option of running cloud functions.
Firebase
Firebase is another popular option and offers a suite of tools for building web apps rapidly. This includes everything from storage, through identity management to cloud functions. Others have mentioned it can get pretty expensive, especially when it comes to their hosting offering.
Virtual machines
Perhaps what was fairly surprising to me was that VMs (virtual machines) are still a very popular option among Indie Hackers.
They require a little bit of upfront setup and ongoing maintenance, but it’s definitely on the cheaper end, which is definitely alluring when bootstrapping. Personally, unless that’s something you’re dealing with daily I wouldn’t recommend going this route – and even if it’s something you’re familiar with I find it that time is better spent on building your offering.
DigitalOcean
Together with Linode and Vultr, DigitalOcean is a very popular option among those that choose to run on their own VMs.
When designing fault-tolerant applications it’s good to bear in mind that their SLA (service-level agreement) only guarantees 99.99% uptime. While it’s the same level major cloud providers offer, DigitalOcean only applies the lost credits 1:1, which may have some implications if you have SLAs for your own customers.
Pros:
- cheap shared VMs
- different regions worldwide
Cons:
- dedicated VMs aren’t that much cheaper than other major cloud providers
- SLA credits policy leaves much to be desired
Linode
Even though Linode, DigitalOcean and Vultr keep competing with one another and their offerings tend to even out pretty quickly after one of them introduces reductions at the point of writing this article Linode has a much better price point for dedicated machines and offer more CPU cores than their competitors.
Similarly to DigitalOcean, Linode has a fairly mediocre SLA, where they refund lost credits at a 1:1 ration.
Pros:
- cheap shared and dedicated VMs
- different regions worldwide
Cons:
- SLA credits policy leaves much to be desired
Vultr
What makes Vultr stand out out of the three cloud providers is their lowest shared tier – starting at $2.5 without IPv4 address or $3.5 with IPv4 included.
Unlike DigitalOcean and Linode, they offer a 100% uptime guarantee and apply credits far more generously.
Pros:
- cheap shared VMs
- different regions worldwide
- compelling SLA
Hetzner
If your project is targeted at EU audience, Hetzner is one of the most cost-effective options out there. With prices starting as low as €3 for a shared VM and €24 for a dedicated one, their offering really is a steal.
Pros:
- very cheap VMs
- very cheap bandwidth
Cons:
- EU-only
AWS / GCP / Azure
With their rich service offering, these three major cloud providers can fit in any of these categories – from basic VMs to platforms such as the AWS Elastic Beanstalk, to serverless functions.
This diversity is both a boon and a curse – while you won’t ever need to leave a cloud provider once you choose it, there’s a good chance you’ll find a significant portion of your time navigating around this complexity, especially in the beginning.
Most notably, AWS (as the first public cloud) has an astounding number of services, tailored to specific use-cases, most of which you’ll never use.
That being said, all of them offer discounts for committed usage of their VMs – something that is lacking in the offering of the other cloud providers. If you can decide how much resources you’re going to need in advance this can bring you considerable savings.
Serverless
Interestingly enough, nobody on the thread mentioned serverless services like AWS Lambda. Personally, I think they’re a great fit for a bootstrapping founder as the less time you’re spending managing your services, the more time you have for growing your product.
I attribute it to lower awareness due to its relatively young age. One major drawback I see here is that the tooling can be quite lacking around serverless and there’s a considerable time investment in terms of the boilerplate required.
That’s why platforms like Firebase, Netlify and Vercel that provide that additional tooling are far more popular, even though they’re essentially leveraging the same building blocks. My prediction is that as tooling in the serverless ecosystem matures, the benefits of those platforms will diminish.
Thanks for tuning in and I hope you are now better equipped to choose the provider for your next project! As always, feel free to leave a comment on Twitter, my DMs are open as well.