8
8 Comments

Scaling the infrastructure for bootstrappers

https://scrumpy.io/blog/growing-a-bootstrapped-startup-with-digital-ocean

We're bootstrappers, so time & money is limited. With DO we're running Scrumpy pretty cheap and stay flexible. ✌️If you have any questions, don't hesitate to ask. 😘

https://scrumpy.io/blog/growing-a-bootstrapped-startup-with-digital-ocean

submitted this linkon June 11, 2019
  1. 1

    this so awesome, thanks for sharing!
    I got to ask, did you choose this architecture over a serverless architecture? and if you did I'd be interested in learning the conclusion you arrived to.
    wouldn't using a serverless infrastructure be even cheaper? I mean recieving a 100k requests every single day would end up costing cents if not a couple bucks, and would cut the need for load balancing and server management

    1. 3

      Load balancing? Why would an early stage startup or bootstrapper worry about that?

      The cheapest DO droplet can easily withstand hitting the top of PH and HN simultaneously. Bigger droplets can handle enormous amounts of traffic. Even Uber stayed off of AWS and similar years into scaling their service.

      For anybody worried they might scale significantly faster than Uber, I'd say: Keep it simple and don't fall prey to Scalability Arrogance!

      1. 1

        I agree.. if you read the post they seem to be planning to integrate a load balancing droplet which is why I mentioned that going serverless would save them that headache should the need ever arise for it

        1. 2

          we had an uptime of 400+ days with our API server, but even a planned maintenance (e. g. to upgrade nginx) results in a few support requests, so people are impacted already.

          we plan to switch to a load balancer setup to be able to shut off one machine without an impact for the users, not for performance.

          1. 2

            At least you're not like Blizzard and taking battle.net down for multiple hours every Tuesday morning and frustring millions of paying users each time 🤣

            Seriously, though, this is totally rational. You'd be fortunate to have scaling problems but reliability is a whole separate concern, especially when your customers have downstream customers.

    2. 1

      not sure about this! we have a serverless architecture for other projects and it's cheaper for development but in most cases more expensive if it's running in production. e. g. managed databases are pretty expensive.

      unpopular opinion: we have experience with running microservices and serverless stuff, but for bootstrapped apps with a small team we go with a monolith.

  2. 1

    This is awesome information. Thank you! I'm currently running on Digital Ocean too. However, since my .NET core webapp is simple. I'm running everything on a single droplet. When I need to scale I'll definitely refer back to this.

    1. 1

      thank you, david! 💕