A Letter to Heroku - Please Tier the Dynos | Hacker News Reader

A Letter to Heroku - Please Tier the Dynos

kent.io

9 pointskentf14 comments

They do seem to prioritize paying customer, particularly during outages.

https://status.heroku.com/incidents/151

"We prioritize getting top-paying customers back online over our larger base of free users, which is why customers (particularly those with dedicated databases) were back online much more quickly than free apps. While we think this prioritization makes sense, we do strive to provide a high level of service to everyone. Even though the outage was much shorter (less than 16 hours in most cases) for our top customers than for our free users (as much as 3 days in some cases), we measure our downtime as the time it took to get 100% of apps back online."

Thanks for the link. Makes me feel better :)

I've never really give heroku a chance.. because of the cheap price of 20$/month for a linode. I once was tempted to use it for a contract but I felt there are so much cheaper and good alternatives..

You get 512mb of Ram for free on Heroku. 1024mb of ram for $36 per month. How much does $20 buy you on linode, and how much is your time worth?

Depends on your skills. Heroku is cheap, but when you start using addons price goes up, in comparison with a vps. I'm a happy Linode user but then again my infrastructure is simple. Spent a few hours configuring everything, but now deploy only takes me a couple of minutes. Good enough for my needs.

What annoys me is that on Heroku, if I need more "Web dynos" or more "Worker dynos", the price climbs really fast. I.e. From that page, http://www.heroku.com/pricing#24-24 it costs 1500$/month... whereas on my Linode, I can do whatever I want, add as much database/workers/web server as I want, and the price is still 20$.

Paying customer here. I hope they don't do that. If you consider a dyno to be a unit of some set of resources, having a free dyno tells you exactly what to expect. Although most apps do not scale as a linear multiple of the number of dynos you have due to other bottlenecks, the resources you have increase linearly. So, having a single unit of resources to try out is a really great model that would give you some peace of mind in scaling.

After all, most successful apps will quickly outgrow their free dyno (and that is without even considering at least an additional dyno for background services).

Free dynos shut off after inactivity (~30 minutes in my experience). So you're probably subsidizing a lot less than you think, since most production apps have scaled beyond the free tier.

Good insight. I still think that having them tiered would make more sense. More curious as to why they aren't.

Customer acquisition, I would think. The cost to them is very small compared to the gains they get by attracting so may people with an awesome, simple-to-use service. Think of it as the "gateway" dyno.

Tiers would just get in the way of the simplicity of their service.

I've seen "tips" on the internet of hooking up pingdom to keep dynes active at all times

Are there particular performance issues you are running into? I've been using Heroku for quite some time and never had any complaints about performance.

You should know that Heroku also automatically shuts down applications running one dyno after a period of inactivity.

Not really. I am really curious as to why they don't. Seems like a natural way to keep them separate in cases there are performance issues that do arise.

I think adding tiers would muddy their service model – the default answer is /always/ not to do something, adding a feature or constraint should only happen with a compelling reason.

Reply on news.ycombinator.com