Razor caching proxy
lindsay at pa.net
Thu Jan 22 15:12:18 GMT 2004
On Wed, 2004-01-21 at 13:53, Julian Field wrote:
> I've just come across this here:
> Has anyone else ever heard of it, or tried it out? I would be interested to
> hear people's opinions of it. I have contacted Vipul directly to ask if
> there is a way of mirroring their servers, and unfortunately the answer is
> no. Apparently the reporting and querying services are very tightly bound
> to each other, so mirrors aren't possible. Shame.
> And what about DCC?
> Is anyone running the DCC daemon to provide their own DCC server.
We run a public dcc server here. We also offered pyzor a public
mirror. It's amazing how little bandwidth and resources these services
> Does this
> help performance noticeably?
After setting up the dccd server, I came to the realization that dcc and
dccd is really cool. If timeouts are your concern, it doesn't look like
you'd every get more than a couple timeouts from one server before the
client chooses the next best on the list of about 16.
If processing time per message is a concern though, it may be a
different story. If you run cdcc info and look at the lowest RTT time,
you can get a rough estimate of the time it takes for a dcc check. At a
glance now, our lowest offsite RTT is 552.18. The local server is
reporting a RTT of 6.74. Thus, I'd suggest we are saving over 500ms per
One thing to note, the dccd server will artificially inflate the RTT
time to a client whom is hitting a server hard. The resulting effect
will be to push busy clients around from server to server as they
continue to pick the lowest RTT time. You may see large RTT times if
you are one of those clients.
> I am worried that too many large sites will
> overload the central DCC servers, when we could run our own.
>From watching the load and network graphs of our public box, I'd suggest
that the dcc server network has plenty of resources available. Our box
is using little bandwidth and is running 99% idle.
> Julian Field
> Professional Support Services at www.MailScanner.biz
> MailScanner thanks transtec Computers for their support
> PGP footprint: EE81 D763 3DB0 0BFD E1DC 7222 11F6 5947 1415 B654
More information about the MailScanner