Hacker News Reader

Comment by inopinatus | Hacker News Reader

All our duelling anecdotes aside, you've picked out the one category I would most definitely not host on AWS: high constant outbound traffic sites. "Don't build your CDN on AWS", I can agree about that. Using public clouds for the spiky part of an otherwise predictable steady-state workload is also a great strategy. Also this:

> Their biggest benefit is to defer the initial setup of a self-managed cluster

And that, I think, is why (nearly) every startup, or indeed every new project, should begin on a public cloud. The capital waste of an incorrect server purchase can ruin a project and makes "fast failure" hard to stomach.

I don't move services away until an alternative is clearly both cheaper in 3yr NPV, and will not constrain future business opportunities. Except in cash cow operations where innovation has ceased, I rate the second criteria more important than the first and a compelling reason to stay on a public cloud.

View on HN

I tend to favour starting with rented servers too, rather than purchases (or leases, rather) - whether that's on AWS, or rented by the month then becomes a much simpler comparison.

As for the CDN, we absolutely agree. I noted elsewhere that even if you host on AWS, if you have any kind of volume of outbound bandwidth use, you should probably get a CDN elsewhere whether or not you think you need a CDN. A good caching CDN can for many people cut their AWS bill dramatically without making you move anything else off AWS.

What about CloudFront?

Traffic pricing for CloudFront is similar to standard traffic pricing on AWS (e.g. for EC2 instances). They charge you also per request and user origin a bit extra, so in some cases maybe even more expensive.

As mentioned, its traffic charges are as high as EC2/S3. CloudFront is useful if your goal is to reduce load on your EC2/S3 setup for reasons mostly other than cost, and want an all-AWS stack.

If cost is your reason for looking at a CDN, or a big part of it, CloudFront will do very little for you unless your pageviews are extremely costly in terms of compute relative to the amount of data returned, and said data is very cache-friendly.

That's very much a niche requirement. To date I've not come across a setup where CloudFront made sense cost-wise.

Good to know. I've noticed that NixOS hosts their binary package caches on CloudFront, which I assumed was because it's cost effective. Those URLs are completely content addressed so they're optimal for CDN caching. Is there a significantly better alternative to S3+CloudFront? I'm pretty influenced by brand names in this situation because there are so many random hosting companies out there.

You can keep S3 if your cache hit rates are good enough and still save a lot. S3 is great for the storage for e.g. durability, as long as you can avoid serving up too much of your content front it.

To be honest I rarely use external CDNs and instead "roll my own" with a variety of providers as with cloud providers + geoip enabled DNS you can get 90% of the benefit at very low rates, but in terms of "brand name recognition" MaxCDN is worth checking. It's not nearly the maximum saving you can get, but it can be substantially cheaper than CloudFront.

Reply on news.ycombinator.com