Folding Forum

Hi!

I'm a student doing a project about folding@home and I have 2 questions about the FAH servers that I haven't been able to find answers to:

What networking technology is used to connect the servers? What is the volume of the data stored in the servers (in bytes on average)?

If someone could point me in the right direction, that would be great. Thanks so much!

Related to the networking technology, check out the publication Folding@home: Lessons From Eight Years of Volunteer Distributed Computing

As for the volume, the last figures I know of are from 2009: http://en.fah-addict.net/articles/artic ... e-home.php. Over 400TB back then.

Here's related statement from Bruce: viewtopic.php?f=15&t=22015&start=15#p222660

The fah client behaves like a web browser. Uses http on port 80.

Project for what class at what school?

I can't think of a reason why the Networking between the various FAH servers would need anything exotic. Exotic networking makes sense between the nodes of a cluster but the FAH servers are not functioning as a cluster.

One or more independent FAH projects are stored on a single Work Server (WS). See http://fah-web.stanford.edu/psummary.html. That Work Server distributes a Work Unit (WU) to a client for processing, waits a few hours or a few days for it to be completed and then accepts the data when the client reports that it has finished the WU. The WS does keep track of which WUs have been distributed to client nodes. If the WU is not returned on time, it gets reissued to somebody else so that 100% of the WUs get completed (Well, almost). As you can see, network communications for all of this activity is strictly between the WS and the Client and standard Internet protocols are used.

There also is an Assignment Server (AS). It's job is to poll the various WSs and maintain a global view of which WSs have active projects with WUs to assign. When a Client needs a new WU, it FIRST contacts the AS which redirects the client to a WS that can provide a fresh WU. Unlike the data that's distributed by the various WS, this is a very simple transaction that gets done once per WU. Again, standard internet protocols are adequate.

There is one more polling operation which is performed by a Stats server. It collects the credit reports from all of the Work Servers and adds that information to the Points Data Base.

Periodically, the data that has been uploaded from the various Clients to each Work Server is analyzed and moved to permanent off-line storage.

There may be some other details that I'm not aware of, but as I said, I don't see the need for any exotic networking technology.

7im wrote:The fah client behaves like a web browser. Uses http on port 80.

Good point. From the Uninstallation FAQ:

From the point of view of the computer, Folding@home acts exactly like a web browser. It downloads files from the internet (with security checksums to ensure that they came from the official site). Indeed, it uses only port 80 (or 8080 if available) and talks to the servers via HTML, just like a web browser. From here, FAH is different: it performs it's scientific calculation and then resends the data back. If you are running Folding@home, you will see network traffic on port 8080 or 80. Only idle CPU time will be used by the client.

This is also reiterated by Dr. Pande in this video: http://castroller.com/Podcasts/FuturesInBiotech/249153

7im wrote:The fah client behaves like a web browser. Uses http on port 80.

That is a very misleading answer. Just because an app uses a web service that uses port 80, does not mean it acts like a web browser. The fah client is nothing like or close to a web browser.

In regards to how the question was asked by the OP (network protocol to servers), a web browser is exactly the correct answer. FAH FAQs and Dr. Pande say the same time.

Yes, the client can be so much more, but not in regards to that question. And Bruce highlighted some of those additional functions/interactions.

See also: What about security issues?

IIRC, F@h uses ports 80 and 8080 because those are best handled by firewalls since browsers use them. Back when the project started in 2000, this was definitely something to consider.

Thank you so much for the explanations and links! I feel like I understand the structure a lot better now. Thank you!
@ 7im, I'm a student at Central CT State taking a Computing and Communication Technology course.