Finding Mirrors for the evolt Browser Archive?

MartinB asks: "I help out running evolt.org, and one of the things we provide is a comprehensive browser archive, with over 100 different browsers, some in multiple platforms and versions, going right back to Mosaic 0.4. This is both a piece of web history, and a resource that lets developers test their sites on browsers which vendors don't offer for download any more. We have an expensive problem - the browser archive chews through 140GB of bandwidth a month and growing, even though we've throttled the FTP server and restricted the maxclients. How do we find people who provide mirrors like these and get browsers.evolt.org spread across lots of hosts?" If you would like to mirror this valuable net resource, please volunteer here (or drop a line to the original submitter)

Comments

This won't help.

[reaper20]

We have an expensive problem - the browser archive chews through 140GB of bandwidth a month and growing, even though we've throttled the FTP server and restricted the maxclients.

For about the next day or too you can count on that getting a little bit more expensive. :)

I hope they find someone, the archive is pretty useful.

Re:This won't help.

[MartinB]

We can cope with a couple of days of /. traffic if it helps us find a long-term host.

Thanks for your kind words about the archive, we'd hate to lose it.

.edu sites

[crow]

Look at where mirrors for things like Linux distributions are found. Many exist at .edu sites. Perhaps you could find a University to help you out.

Get a dedicated at RackShack

[Anonymous Coward]

They often have $99/mo specials, and include 400 gigs of bandwidth. That's enough you could have your 140gb covered, AND still sell/use the remaining resources for something else. So, yeah, find someone else(s) that need 260 gigs, and split the cost with them (or bill them higher and cover your costs).

Long-Term Solution: P2P

[crow]

What we need is a good P2P system for distributing software. Something designed for distributing software packages. I'm not sure how well this would work for the example in the question, but imagine the following system:

The web site that offers the software would use a link like:

p2p://www.mysite.org/p2p/package7.xml

The above file would be an XML file describing a given package. It would be downloaded using http, but then automatically go and start using the resulting file to search for and download the actual package. The XML file would give the MD5 sum for the entire package, as well as for each 1MB chunk. A P2P network would be used to search for and download chunks based on their MD5 sum. The client would verify the sums and reconstruct the package.

Now all you need to do is get some site to share some of your more popular files.

Re:Long-Term Solution: P2P

[mlinksva]

P2P is the way to go. Indentifying files by hash is the way to go. A colleague of mine has a proposal for generic browser/P2P client integration.

ibiblio?

[_vapor]

Have you considered ibiblio.org? This seems right up their alley. They have a public FTP archive that can handle a lot of traffic, and they mirror Linux distributions, among other things.

funet

[photon317]

I don't how easy or hard it is to get a mirror from them - but ftp.funet.fi has always been my favorite high speed mirror. It has been the most stable, reliable, and high-speed mirror I've used for downloading various stuff over the past 8 years or so, and they mirror *tons* of useful things on the net.

Web Designers

[Sauron23]

Web page designers and web programmers would likely be two groups that could use access to older browsers for compatibility reasons. Someone already mentioned .edu's as a likely candidate for a mirror. Consider contacting their media labs in addition to the CS departments. Look for schools that offer courses in webpage design in their College of Arts.

UK Mirror Service

[Spanshter]

The UK Mirror Service (http://www.mirror.ac.uk for the paranoid) provides a lot of useful mirrors, including a nice tucows mirror.

They will probably mirror something like this.

My recommendation: eDonkey2000

[Lenolium]

My recommendation to you is to use eDonkey2000, use P2P for what P2P is supposed to do. eDonkey2000 uniquely identifies each and every file on the network with a MD4 sum and a filesize. You could just put a bunch of ed2k:// links on your site to replace the current downloads, and make sure the server is running edonkey2000 and serving up those files (So that there always is at least one good place to get the files, even though edonkey2000 will go for multiple hosts with the same file at the same time). To even be more helpful, you could always connect to a specific listing server, and give people that address. At 140GB/month, you could actually bring a good amonut of legimiate traffic to this P2P network.

EDonkey Homepage

ML donkey homepage (my preferred Linux client)

Re:My recommendation: eDonkey2000

[0x0d0a]

I like the P2P idea, but with a different network.

Put out the vast majority of the archive *only* on Freenet. A nice big distributed caching proxy network, browsable with a web browser, and growing in content every day....and it could use a valuable, exclusive resource like your own.

Legality

[duffbeer703]

You may have difficulty finding a mirror willing to face possible heat from whatever vendors produced the software.

I've noticed that alot of free-to-download software comes with EULA restrictions prohibiting you from redistributing them. I'm not sure about old browsers, but I would hardly be suprised if you are probhibited from disributing them, even though they are abandonware.

Re:Let the wayback machine archive it

[Kizzle]

Thewayback machine only archives text and graphics.

Cheaper Hosting

[Rayonic]

Chew up a lot of bandwidth? I hear SA Net Hosting offers cheap hosting - 24 cents per gig, supposedly.

Sure, you might be their first customer, but they seem to be on the up-and-up. Look here for more info: http://www.sanethosting.com/.

Charge for the bandwidth

[unitron]

If people came to you and said "How about letting me have a copy of foobrowser 0.x on CD", you might not charge them for the software itself, but you wouldn't feel bad about making them pay for the CD itself, would you? Or for telling them to provide a blank CD-R? Why not set up to collect a Paypal'ed bandwidth fee, or get them to mirror what they download for a few days or over a throttled connection?

Supporting free speech doesn't mean that you're obligated to go into your own pocket to provide everyone with free beer.