I run my blog off an EC2 Micro instance running Apache, which Amazon conveniently offers a free year of service for. (And it's not expensive paying for it yourself either ($69/yr for a Small). I used to use HostGator, AWS is much nicer--apart from outgoing emails, but once you're set up you're good.)
Pingdom is kind of nifty for its comparisons. I got the author's site to be only in the top 97% by running the benchmark again. My own site varies from being in the top 90% to 96% (I did get a 99% once), the only real "optimization" I do is reading cached pages from "disk" (EBS is notoriously pretty slow). I could probably shave off an average 100ms with some really simple tricks. Under heavy load I'm certain I'd fall over on such a puny machine that isn't even using nginx+memcached.
According to the pricing page [1], a reserved Small is $69/yr plus $0.039 for any hour the instance is running, which could add up to another $341 for the year.
Good catch, I just quickly glanced at the figure without remembering the hourly rates still apply. Going with the yearly-reserved Micro you're at $128 for the year, which is a bit more than many shared hosts but at least you have absolute power over the software.
Just for a Jekyll blog, you could just host it on a free Heroku instance with an Nginx server and put CloudFlare in front of it.
In a free Heroku instance you have a lot of juice actually, especially for an Nginx server, the only problem being that the instance will go in idle mode after an hour of inactivity and so unlucky visitors can get some latency on the first request, although it's not that awful.
On the other hand CloudFlare is a pretty good proxy that works like a CDN, so with the right caching headers set, CloudFlare will serve many requests from its own cache.
> Pingdom is kind of nifty for its comparisons. I got the author's site to be only in the top 97% by running the benchmark again. My own site varies from being in the top 90% to 96% (I did get a 99% once)
If you strive for geek cred, you can get
> Your website is faster than 100% of all tested websites
The trick is to 1. have a static website, obviously, and everything inlined (so basically a single index.html); 2. run the test twice — dns responses should cache during the first one; 3. host your website in Amsterdam which is where they have (at least one of) their testers:
> Tested from Amsterdam, Netherlands on October 16 at 19:04:21
Pingdom is kind of nifty for its comparisons. I got the author's site to be only in the top 97% by running the benchmark again. My own site varies from being in the top 90% to 96% (I did get a 99% once), the only real "optimization" I do is reading cached pages from "disk" (EBS is notoriously pretty slow). I could probably shave off an average 100ms with some really simple tricks. Under heavy load I'm certain I'd fall over on such a puny machine that isn't even using nginx+memcached.