Alex,
I have some questions regarding polygraph 2.5.0 and polymix-3:
- Do I have to upgrade to FreeBSD 3.4?
- If I don't have to upgrade, are there any changes that I have to do to my
current FreeBSD 3.3 machines that ran polymix-2 on polygraph 2.2.9?
- I noticed that polymix-3 has the fill-cache phase included. Am I correct
to assume that this means that the working set is full by the time top1 is
entered? if so,
what is the need for top2 (which is another 5.5 hours)?
Thanks, Ittai
-----Original Message-----
From: Alex Rousskov [mailto:rousskov@ircache.net]
Sent: Tuesday, June 20, 2000 1:14 AM
To: polygraph@ircache.net
Subject: Poly 2.5.0 available
Hi there,
Polygraph 2.5.0 is available at http://polygraph.ircache.net/sources/
Today is the deadline for releasing a beta version of the Polygraph
code for the third "cache-off". The "milestone" page describing the
current status is at http://bakeoff.ircache.net/N03/miles/june19.html. A
HOWTO preparation document will be available shortly. Please check those
out if you are considering participating in the next test. You are
required to start testing and submit complaints within the next four
weeks.
This version is a major release incorporating most of the features
we have been working on since the last bake-off. I will briefly mention
some of them below. A few other major enhancements will be committed and
announced soon.
A new "agent", called polypxy has been added. Polypxy is a simple
implementation of a "fake" caching proxy that can talk both HTTP and
ICP. Polypxy can be pro-active. That is, it can emit request stream just
as polyclt. In fact, polypxy can be viewed as a merger between polyclt
and polysrv with a fake cache in between. The released version of polypxy
can work in "sibling mode" only. That is, it will refuse to serve misses
to peers. We are working on the "parent mode" functionality.
Another long-awaited feature is support for file name "extensions"
in URLs. We have been using that feature for internal L4+ switch tests,
and it is now committed into the public release. Note that Content-Type
header fields are also supported.
Version 2.5.0 supports PolyMix-3 workload, including such new
features as "hot spots" simulation, "reload" and IMS requests, fill
phase integrated into the performance test, and better coordination
among robots and servers.
Polygraph code structure has changed significantly to address growth
needs as well as portability concerns. As a part of that change, Polygraph
now partially supports "make install" target and can probably be built
outside of the source tree, even if your maker does not support VPATH.
This release is definitely a "beta" software. We have not yet tested
this version outside of the FreeBSD environment. A sizeable list of
portability patches, including native Win2K (MS Visual C++) and HP-UX
(aCC) support, await their inclusion. We did not forget about a few
Linux and Solaris bugs that needed to be fixed (and there are probably
new ones). Their time will come soon.
The following big items are on our TODO list:
- Portability fixes
- Improved run automation tools; make it easier to start,
stop, and monitor a test
- Improved report generation tools
- Documentation synchronization, Web site improvements
We also realize the need for improving PGL, but that may have to wait
until September. Eventually, PGL will become somewhat similar to Perl:
an interpreter of an intermediate code form, allowing for run-time
cooperation among language components (as opposed to the current static
"configure then run" approach).
Enjoy,
Alex.
[ bbc-ed to bakeoff@ircache.net ]
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:13 MDT