|
Lars, Alan Robertson, Joseph
First off, overall I think the Press relaease looks good, and I'm very
pleased to see that you listed us as a user of the software.
This:
> JANET Web Cache System, the UK national cache service, clustering 52
> servers into a virtual proxy server, (yes, that is fifty-two)
Is * WILDY * inaccurate. This is *potential* possibility at some point
early next year.
It is ** N O T ** the case now. Currently we have the service under trial,
and I have had private communications with Wensong to say that things are
looking very good, but we are nowhere near that sort of situation yet.
I wish I'd been consulted (only for accuracy!) before you'd used this
particular piece of information - I would've been able to give you with
impressive figures for how the testing is going - things like 7 Million (*)
proxy requests per day, load balanced within 0.5% of optimum, >60Gb per
day, 100% uptime on the load balancer since trials started, that sort of
thing, and it would've been accurate and defensible!
For various reasons I cannot disclose to the world at large yet, I'm
producing a document stating what hardware we do have in the cluster, and
to quote:
> The system currently consists of 1 Director and 4 Real Servers. These
> are all P400s with 1/2Gb of RAM, and 50Gb of disk.
*Not* insubstantial, but likewise, not 52 servers!
On a personal note, if things continue as well as they have done, then we
may end up in that situation.
If Lars, Alan, Wensong, or Joseph want to discuss this with me (or
anything related with the virtual servers stuff :-) - *please* feel free
to contact me - in the meantime, please amend the web page with the above
(still glowing I hope!) notes on traffic through the cluster.
On a final note though, so this email doesn't come across as all negative
- the software is bearing up *extremely* well in a very harsh production
environment - and I for one continue to be impressed :-)
Also, aside from the inaccuracy, we're proud to be associated with this,
and I personally think the press release looks very good.
Regarding this point on the page:
> In the future, we'll have better admin tools,
I'm working on a framework of bring all of our various monitoring tools
into use, so that we can maximise uptime, and still configure the system
simply - amongst which includes a nice pretty monitor screen that's
suprisingly useful, and also a router/switch style configuration console.
One thing I'm also hoping to implement is an automatic system for allocating
weights to servers based on data from the monitoring subsystems - so for
example the system would automatically detect when given 3 servers which
ones become overloaded at what levels, and assign weights accordingly.
Naturally all this stuff will be released back to the community.
Michael.
--
National & Local Web Cache Support R: G117
Manchester Computing T: 0161 275 7195
University of Manchester F: 0161 275 6040
Manchester UK M13 9PL M: Michael.Sparks@xxxxxxxxxxxxxxx
On Fri, 12 Nov 1999, Lars Marowsky-Bree wrote:
> Good Morning,
>
> together with Wensong, Alan Robertson and Joseph, I wrote a press release /
> announcement, which has today been submitted to Linux Today, Freshmeat and
> Slashdot.
>
> Please find the announcement at
> http://www.LinuxVirtualServer.org/lmb/LVS-Announce.html .
>
> Sincerely,
> Lars Marowsky-Brée
>
> --
> Lars Marowsky-Brée
> Network Management
>
> teuto.net Netzdienste GmbH
>
> ----------------------------------------------------------------------
> LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> To unsubscribe, e-mail: lvs-users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxx
> For additional commands, e-mail: lvs-users-help@xxxxxxxxxxxxxxxxxxxxxx
>
----------------------------------------------------------------------
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
To unsubscribe, e-mail: lvs-users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxx
For additional commands, e-mail: lvs-users-help@xxxxxxxxxxxxxxxxxxxxxx
|
