The Coral Content Distribution Network | Hacker News Reader

The Coral Content Distribution Network

coralcdn.org

144 pointshrjet43 comments

Note that they don't cache anything over ~50mb, and it's a relatively small cache (~4GB nodes, cleared LRU first), in addition to only revalidating every 5 minutes.

http://wiki.coralcdn.org/faq.html

So don't use it for big files or anything that changes more than every 5 minutes.

Useful to know, but it's still an easy and efficient way to reduce load from static elements.

It's particularly useful when an article or something you want to read gets slashdotted.

Why don't forums like Hacker News, Reddit, etc. build this into their submission process, so the article/page is cached upon submission, and readers are directed straight to the cached copy?

bump

And don't forget, if you're outside the US, appending their CDN domain to yours brings your static content (and the browsing trails of your users) silently into US jurisdiction. Or at least that's the DHS and FBI interpretation of the law with respect to .com, .org, .net

Browsing trails, I see that.

But content? Anyone can make a .net domain and mirror non-US content on it. Does that give DHS/FBI any legal power over said content? Not over the original, at least.

.org is not controlled by Verisign so it's harder for them to take it down. Also they can't control the content, just the domain.

Isn't this very old? I am sure I stumbled upon this while doing some research on a CDN project I was interested in about 5-years ago.

I would assume that CDN technology has progressed significantly since then.

The impression I got at the time, was that it wasn't maintained - as well as I would have liked. I could be mistaken, but that was the impression I got a few years ago.

From the frontpage:

»CoralCDN has been continuously operated since March 2004, running on 300-400 servers on the PlanetLab testbed, spread worldwide. As of 2011, it receives 25-50 million requests per day from a few million unique clients.«

In the beginning you had to add .nyud.net:8080 but it changed to the more convenient .nyud.net along the way.

From their overview page:

"20 Aug 2012: We're still here! While active development has been stopped for a while, we continue to operate CoralCDN as an open, free service. It's now been running continuously for more than 8 years (since March 2004), and continues to get a few million users per day at last check. Enjoy continuing to use the service!"

I thought everyone knew about this? I use it for mirrors when a site looks like it's going to go down. Just append .nyud.net to the domain and it gets cached.

Neat trick!

Same here. CoralCDN has been around for ages, I've used it to auto-mirror images and other static content from sites to reduce bandwidth consumption. Works really well.

IIRC, it was originally devised to fix sites melting down under traffic spikes generated by Slashdot browsing crowd.

That's what I used it for! Of course, I never published anything worth getting Slashdotted :)

And to grab US only content from the UK. Ironically, it's usually BBC content...

CoralCDN is old an unmaintained.

For the new hotness (web only), try https://peercdn.com/ :)

This requires Javascript though, and its operation is way more complicated. It depends on peer-to-peer delivery, which is pretty hard when most users are behind NAT'ed routers and proxies. Furthermore, you need quite a few concurrent users requesting the same resources before it will even start using this CDN.

Finally, it's "currently in private beta", which means it's relatively untested in the real world, and not open like Coral at all.

I think there's just no comparing the two.

Yes, but it's the new hotness :)

Much like using Hadoop on 15MB CSV files? :)

15MB of Big Data.

It would be nice to have a dropin JS library for CoralCDN too.

Great project. For anyone interested in serving up static resources seamlessly, you can use something like the following in an Apache config:

  <IfModule mod_rewrite.c>
    RewriteEngine on
    RewriteCond %{HTTP_USER_AGENT} !^CoralWebPrx
    RewriteRule "\.(png|jpg|css|js)$" http://%{HTTP_HOST}.nyud.net%{REQUEST_URI} [L]
  </IfModule>

I wonder if this can be used in any interesting ways as a drop-in performance enhancement. Using it as an automatic domain sharding solution might be viable?

it is unmaintained. to quote a mail from january 2012 (the maybe twentieth latest mail in the -dev mailing list):

    > CoralCDN is basically a "one-person job", and I never got around to 
    > building a great monitoring-and-notification system. 
    > In this particular 
    > case, I didn't have as great access to email as I usually do.

coralcdn.org is blocked by HR policies where I work; thus I would recommend against using it for business purposes.

Same for me, both this domain and nyud.net are blocked because they're classified as proxies.

I am quite sure the proper way to resolve this problem would be to change the said policies.

I think the thing is that you cant change policies on your users companies

But users can also have arbitrarily horrible policies that will restrict any possible thing you might like to do. No images loaded, for example.

It is also blocked by the "Low" setting of the OpenDNS content filter, a fairly popular service and setting.

Well, you can use it to trivially get around all other blocks. Points more to the futility of trying to block "bad" sites name by name than anything else.

For some reason the DNS servers my Swedish ISP provides seems to ignore/block all lookups for anything containing nyud.net. I have never encountered anything at all being blocked before with my ISP, so it seems a bit strange. Too bad if other ISPs also are blocking CoralCDN. I tried Googles DNS server (8.8.8.8) and everything works fine with that.

Who pay all the servers? Or how can anybody contribute?

Everybody pays. It's operated using PlanetLab, which is run by over 500 universities from all around the world. You can't contribute to the operation, other than maybe funding a PlanetLab node somewhere.

Ok thx, i misunderstand PlanetLab, thought it was a company ;)

What the heck is that "Illuminati" thing at the bottom of the page mean?

http://web.archive.org/web/20100723133654/http://illuminati....

And now I know about this? I should hit me in the face.

It hasn't been very fast or reliable the few times I tried it. I'd rather pay for a decent offering.

Time out.

I just tried the site and it worked okay for me (2mins after you posted this). Also the CDN test seems to work okay as well. Try doing a trace route.

Try appending .nyud.net to the URL.

Reply on news.ycombinator.com