Project Status

03/03/97
AIX collector potentially complete.  Work on network tracing completed.
DEC has still done basically nothing to impliment the AIX api into the
DI.  Roger Grant says that DEC is paring down the CP team.

01/08/97
DEC has promised again to work on the DI code.  
10/01/96 
DEC has been given the AIX collector code, waiting for the DI work 
to be done.

08/09/96
Gee Pang has finished the port of the DI to AIX. He now works
for Microsoft.  

Work on the AIX collector is progressing, network tracing yet to be done.

01/16/96

DEC will be visiting on 1/19 to discuss the details of the project.
A general presentation of the DEC Capacity Planner software will be
scheduled for 2/6 during the AST lecture time slot.

11/28/95

Jim is setting up a presentation by BGS (Best/1) for Dec 18. 

11/15/95

Jim is looking into licensing the Best/1 UNIX performance benchmarking and
modeling tool to assist in benchmarking amd modeling clusters.

Jim 

Date: Tue, 14 Nov 1995 08:57:49 -0800 (PST)
From: Ellen Jensen 
To: Doug Luft 
Subject: Re: Benchmarking denali

Doug,
It's nice to hear about progress on the benchmark of linking.  Is there
a description of what you have done and how it will work?  Will we get 
reports automatically or will you be doing some interpretation.

I know Jim has talked about continuous monitoring of the loads on denali
and THurgood.  Will this be working as well?
Thanks
--Ellen

On Mon, 13 Nov 1995, Geri Bunker wrote:

> Doug,
> Thanks for creating the "tpx" software on libsrv1 for benchmarking denali 
> performance.  I've met with Jill and Bill from the Serials Linking 
> Project, and we discussed how we might use your program to test various 
> things on denali.  
> 
> At this point, we're looking at a 2-phase test, with the first phase being
> run while the Libraries are closed.  From your description of the program
> and what will happen on denali, we're thinking that Phase I should be run
> by someone who can simultaneously run a root shell on denali, just in
> case.  (We'll be happy to suggest the program parameters, but we'd also
> like to get some feedback from the IS shop before we actually run the
> tests.)  
> 
> Phase II will require close scheduling with the Libraries, not only so
> that we can be prepared for slowdowns, but also so as to have folks
> participate in data gathering at the staff (circ and tech processing)
> desktops.  E.g., we need to observe what happens during interactive circ 
> transaction processing as well as during the submission of regular 
> "batch" jobs to denali's single transaction queue.  
> 
> Perhaps we can move this process forward by asking for comments at this 
> point--does this approach sound like it will uncover the critical factors 
> we need to test?  Thanks.
> Geri
>