mersenneforum.org  

Go Back   mersenneforum.org > Fun Stuff > Lounge

Reply
 
Thread Tools
Old 2003-12-03, 01:06   #1
GP2
 
GP2's Avatar
 
Sep 2003

2,579 Posts
Default The difference between P2P and distributed computing and grid computing

The theinquirer.net article refers to PrimeNet as P2P software... that's incorrect, it's standard client-server. Clients don't talk to one another.

The Cnet article equates (volunteer) distributed computing with grid or utility computing. That's also inaccurate: the architecture is entirely different.

With grid computing, the computing resources are available through a grid service. In other words, computing is done on the server (or its back-end). Clients "plug in" to the service and consume it in much the same way that a utility customer plugs into the electrical grid and consumes electricity. Or at least that's the concept... it might turn out to be another hot buzzword that never really catches on.

With volunteer distributed computing, the computation is all done in the clients, who connect to the server at irregular intervals to upload their results. Basically the architecture is inside-out compared to grid utility computing.

Thus the standard OGSA architecture for grid computing really doesn't apply to volunteer distributed computing efforts. Some tentative efforts have been made to create a standard platform for volunteer distributed computing, such as BOINC, but right now each project uses its own infrastructure.

Now mind you, you can have a virtual grid service whose back-end consists of subcontracting out the computational work to thousands of home PC users (for money). That would sort of be a combination of the two architectures. But most grid services talked up by the major computer manufacturers would use that's company's in-house computing resources.


Another way that volunteer distributed computing projects differ is because of the lack of central command authority and the need for recruitment and retention of participants. In other words, there's a "sales and marketing" aspect in addition to the purely technological aspect, so a technical computational solution is only half a solution.

For instance, it's absolutely essential to devise a metric to give credit for CPU time spent, which must be fair, accurate, and cheater-proof. Participants' rankings according to this metric must be available in an elaborate online "stats" scoreboard. This metric is more or less cast in stone, because zeroing out the scoreboard and starting over will cause many participants to quit the project. The metric must take care to incentivize the right participant behavior and not create a conflict of interest between the project leadership and the volunteer participants, because most project participants are caught up in the competitive aspect and if there is a conflict between goals of the project (scientific benefits) and maximizing their own score, many participants will choose the latter.

It's necessary to allow users to participate as individuals or part of a team. It's often useful to create a screensaver version of the client program in addition to a command-line version, display some aesthetically pleasing albeit meaningless graphic. It's necessary to respond promptly to special user requests: joining and leaving teams, merging two teams into one, etc. etc.

And so forth.
GP2 is offline   Reply With Quote
Old 2003-12-03, 09:38   #2
Prime Monster
 
Prime Monster's Avatar
 
Aug 2002

22·5·13 Posts
Default

The the Inquirer article has been corrected

PM
Prime Monster is offline   Reply With Quote
Old 2003-12-03, 14:13   #3
Old man PrimeNet
 
Old man PrimeNet's Avatar
 
Jan 2003
Altitude>12,500 MSL

101 Posts
Default

Entropia was often mixed up with the P2P space (sometimes by the company's own PR), so I'm not surprised they dug this up again in relation to GIMPS.

The press are a wiley bunch, often friend, sometimes foe. Most are sharp professionals. It's important to work with them but roll with the punches - it is, after all, their news channels. Generally, the press is more accurate if we produce a focused, well-written release and stay on message so they can use it mostly as-is or in quotable bites specifically written in for that purpose.

In this press, most of the articles stay nicely on message, directly quoting our press release. Those that seem off-mark are typically summaries of other and related articles - the second-hand retelling phenomenon.
Old man PrimeNet is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Etiquettes of Distributed Computing a1call Miscellaneous Math 8 2018-05-21 16:25
Considering getting back into distributed computing jasong jasong 7 2016-03-28 14:36
GNFS distributed computing andi314 Factoring 50 2005-05-11 00:56
Distributed Computing Survey garo Lounge 11 2004-09-01 03:31
P-1 factoring as distributed computing jocelynl Math 2 2002-11-23 00:27

All times are UTC. The time now is 17:54.

Sat Aug 8 17:54:57 UTC 2020 up 22 days, 13:41, 2 users, load averages: 1.84, 1.83, 1.73

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.