heruko performance vs shared hosting performance

This guy is using godaddy shared hosting for his wordpress blog. I think it's currently $5 a month. I was thinking to use Rails for his next project. But I wouldn't want costs to exceed more than $20 per month. I contacted heroku and they said that 1 dyno is free to host your ruby on rails app and each additional dyno is $30 a month. A single Heroku dyno provides 512mb ram and 1024mb swap. It has 4 (virtual) CPU cores (Intel Xeon X5550 @ 2.67GHz). 512mb ram doesn't seem to be much, particularly for memory-intensive processing. Although, the multiple CPU cores may enhance multi-threading. That said, this really isn't going to be a memory hog. Now I was curious to compare these specs with that of a shared hosting environment (where many different customers share the same resources). For example, godaddy says this about their shared hosting "Currently there is not a ‘set number’ for CPU resources. Mostly because we have multiple hardware platforms in the wild. Simply put, if your processes degrade performance for other customers on the server(s) we will take actions to protect the other customers." So that's not much of a comparison. But does anyone have any experience on shared hosting (I know that most shared hosting environments dont support ruby - but Im sure you used it for other platfroms like asp.net/php) vs heroku, and if someone is used to shared hosting with moderate traffic to their site, would you notice a significant performance downgrade if the site is moved to heroku on 1 dyno?

Heroku is prohibitively expensive, and GoDaddy is the Wal-Mart of web hosts.

Why not use WebFaction? It’s much better than GoDaddy and much cheaper than Heroku.

would you notice a significant performance downgrade if the site is

moved to heroku on 1 dyno?

Given that the heroku entry tier is free and super easy to set up, why not just deploy an existing app you're familiar with and compare the performance yourself?

Or develop the new app, deploying to heroku as you go, and see if the performance of *that specific app* -- which is the only thing that really counts -- is acceptable. If not, pay to upgrade or move on :slight_smile: