Mib Software Newsrate Usenet Server Comparisons


I peer with just a small number of other sites. Is this bad?


> I wonder how number of peers plays into performance.
> 

With good peers, less than most people imagine, I'd bet.

You should only need 3 good peers. (By "good" I mean able to transfer 80% of a feed to you, if all the other feeds were turned off.)

Anything above that might be seen as a waste of CPU and bandwidth, because you are going to be getting more incoming connections, with more articles offerred that get refused with "Nope, I already have that article."

Even if you only have one or two peers, you can still have a "good" server, or even "excellent" server in a small subset of newsgroups.

newsrAte reports can be used to evaluate and balance peers. When you get a history of statistics (say like two weeks worth) turn off a peer for 12 hours. You man not even notice the difference in the next day's report, if you have enough other "good peers."

If you have the hardware, you can dedicate a server to receive only articles from one peer. This server then forwards the rest to the other servers in your system. If newsrAte is run on that one dedicated machine, you will in effect be measuring how well your peer does. You will be taking every article offerred from them. (Assuming you don't backlog, etc.)

More Detailed Topics
Why do some sites decide to not carry binaries or alt.*?
We only have a T1. How can we be doing 100%?
What is considered "excellent" or "good" or "poor" performance?

Overview and Related Topics

How do I signup?

Up to The newsrAte RKT
Up to newsrAte home
Up to Mib Software home

Copyright 1998, Forrest J. Cavalier III, Mib Software
INN customization and consulting
Email: newsrate@mibsoftware.com
Voice: 570-992-8824