Is Shelly Cloud faster than Heroku? | Hacker News Reader

Is Shelly Cloud faster than Heroku?

blog.shellycloud.com

30 pointsmribica23 comments

Could someone that has used shelly cloud in production share their thoughts on stability and service of it please?

Not specifically about this post or topic, but I kind of miss that blog posts don't have the Conclusions...call them summaries, if you will..at the start of a post.

I get it, all authors want all readers to read the whole of their masterpiece. Unfortunately, though, we live in a world where we are deluged with fluff and rubbish; we need concise, detailed, and conclusive summaries at the start in order to make a judgement as to whether reading the details is necessary or even valuable.

I don't think trying to hold out out of fear that the whole post won't be read is all that productive; possibly even counter-productive.

I agree. One of the things I love about scientific papers is being able to read the abstract to determine if the whole paper is worth reading.

I know that bashing Heroku is popular this days but seriously i don't like what you did.

I'm not a fan of your way of getting users by pointing some non-objective figures and saying hey we're faster & better than the competition.

Seriously, you should benchmark and say "we can do this, this and this", and let your users decide if you're faster, better.. because doing it in your way, is just ugly and not fair.

Btw, I like your design. You mention redis on your site, but there is no redis in the pricing.

I like it.

They offered way more information to us than the route you are suggesting and opened themselves up to more internet scrutiny as a result. I know that "Shelly is faster than Heroku says Shelly." isn't the first thing you want to hear but they offered a comparison in key areas that I find interesting.

What if they just took all the numbers that Heroku offered, then offered their numbers in all of those areas. That is essentially what has happened but we don't need to compile both sets together ourselves because it was already paid for someone to do it.

Arguably Shelly offering numbers that could be compared with Heroku's numbers would be nearly the same thing but less convenient.

To run redis, you only need to pay for the virtual server and storage, that's why it's not mentioned separately in pricing. You can run all non-replicated databases (postgres, mongodb and redis) this way.

Are the response times faster because the company and its servers are both based in the EU? So, really the response time charts are showing the benefit of simply eliminating a cross Atlantic network hop? So, is this a solution for companies targeting EU customers?

The tests were ran from a Hetzner server against Shelly Cloud and Heroku EU, so that shouldn't affect the results.

We are however targeting EU customers, we have servers in EU datacenters. We have plans to expand to the US later.

Thank you for the response!

I'll keep an eye out for your US launch! Competition is a good thing.

Heroku Europe[1] is a better solution in that case.

[1] https://blog.heroku.com/archives/2013/4/24/europe-region

There is not much of Atlantic between Germany and Ireland ;)

Have you considered data centers located in North America at some point?

We're planning to do this eventually, but we're focusing on EU right now.

Doesn't matter, only supports Ruby on Rails.

Might care when it supports stuff I actually use, like Heroku does.

We support all Ruby frameworks, not only Rails.

We believe that by focusing on Ruby we can provide better service than we would if we had to support multiple runtimes and application architectures.

My node.js setup wants to thank you for your attention to our kind. I think I'm also hearing some Django persons hissing on this.

But thanks, reading your "We are so much better than Heroku because X" auto-congratulatory post to then see you stumble on the issue of multiple languages was funny :)

We're not saying that Shelly Cloud better than Heroku in general, we're just pointing out that focusing on one language has it's benefits. There is a market for generic solutions like Heroku, and specific ones like Shelly Cloud, Nodejitsu, etc.

Shelly cloud is an outgrowth of a application development company with extensive Ruby and Rails experience, and probably limited resources.

It makes sense for them to focus on what they do best rather than a broader, so-so service.

Money quote: "If someone doesn't really hate your product, it's mediocre."

Well, I have an installation on Shelly and it's not Rails app, but Rack based one - Sinatra. So not only Rails are supported.

I would only take seriously tests like this if my apps consisted of displaying "Hello World" to paying customers. So far I couldn't find a business model to monetize this type of application.

As a result, I find these tests to be meaningless no matter how pretty the graphs are.

It is customary to reply to any benchmark game involving libraries and framework with "it's too simple; it's meaningless".

But would it be meaningless if you tried to gauge the differences in bandwidth and latency of two datacenters (e.g. provided by the same company) ?

Would it be meaningless if you tried to gauge the effect of a dumb vs. an intelligent routing mesh ?

Maybe it is meaningless, but you don't have the data to back that claim. At least the post provided the code so you can try for yourself.

By keeping the test apps simple, we limit the app code's effect on the results, and show the overhead of the providers' infrastructure.

You're welcome to fork https://github.com/shellycloud/baseline-performance and provide a better test.

No thank you, I have other work to do now. That reminds me I have to drop HN and get back to work. So excuse my short reply.

An even better way to test your platform would be to offer future customers free accounts with reasonable limits to test their applications with real data and load. I realize you offer 20€ for free, but that doesn't cover a real test.

Reply on news.ycombinator.com