|
|
Enron Mail |
Steve,
Thanks a lot. I think that having the pseudo code will go a long way towards understanding how the system works and making sure that there are no bugs in translation of a business problem (for example, complicated credit insurance deals with multiple triggers and conditionality) into the code. Regarding Tanya's attitude. Just a few points. 1. I don't think she has the skills to do the system administrator's work and she does not have the necessary privileges. This explains why she keeps asking Winston for help. It's not that the work is beneath her. 2. Some members of Tanya's team came to me complaining about Winston. He effectively told them to go away and work on the "research projects" and that he would take care of the IT issues. I don't think that it's just Tanya's issue, though I agree that a more outgoing personality would be helpful. 3. The reality of this situation is that the internal customers beat on Tanya and me whenever there is any performance problems and/or they intuitively disagree with the results of a run. They could not care less about the demarcation line between IT and Research. They also want Tanya to sign off on the model and she cannot do it without full access to the code. The bottom line is that we are in full agreement: Tanya and Winston have to work as a team and I shall work on my end to make sure that it happens. Credit is emerging as a critical issue for Enron for the next few weeks and the system cannot fail. Vince From: Stephen Stock/ENRON@enronXgate on 01/11/2001 08:23 AM To: Vince J Kaminski/HOU/ECT@ECT cc: Subject: Progress Vince, I got feedback from the lunchtime research meeting that you were talking about some specific solutions to performance of IT systems... In particular distributed processing. Also I heard that you had concerns about the use of multiple languages etc.... Both of these sound like what I was discussing with you on previous occasions... Do you feel the need to discuss these further? The multi-language issue isn't really that much of an issue, as the current system is 98% java right now. Although I am a big fan of C/C++(it is my main development skill) , I am also very aware that Java is a much more evolved and robust language. I had serious doubts about the performance, but I've had a review conducted, and the results are showing the Sun Unix implementation to be nearly as fast and in some cases faster than C/C++ because of something they call Hot-Spot technology. (its an instruction caching technique, I believe). The concerns I expressed to you, were really about how technical people justify the use of a language on the strength of a relatively meaningless metric like portability. On the issue of distributed processing... the original review I had conducted by our architecture group pointed to that as a solution, and as Zhiyong Wei is already working on Global Valuation project, Winston is actively working with Zhiyong to see if he can model the VaR architecture on that, and also to find a common Valuation piece between the systems. I'd like the opportunity to talk to you about these issues if you have some time over the next few days? Also, I sat in on the Tanya / Winston meeting yesterday and as per our discussion at the elevator, I attempted to help her argument by suggesting to all present that she was trying to perform triage on the code... I.e. Seperating research domain problems from IT problems. She said that stepping through code was the only real way in which she could get a feel for where performance bottlenecks were. I asked her how she would measure that, and she said she would instrument the code manually by inserting timing elements at strategic points. I mentioned that a profiling tool could probably do this job for her. Tanya again said that stepping through code is the only way she can get an idea of the code, and that studying documentation wasn't enough. About 6 weeks ago, I commissioned a team to document the system down to psuedo-code level and will be able to provide this to you and your team soon. (in fact I've asked for a draft copy to be given to Tanya right now), and Winston is also working on a draft Research/IT "working together" document, which will identify how the exchange of information takes place. Tanya also gave the impression that she wants a dedicated IT developer to do all the environment setup for her, because she doesn't really want to have to do that. I think that this is probably the root cause of the issue. The IT guys are working very hard and her handling of the situation is not good, as it gives the impression that this kind of work is beneath her. She is claiming that they are un-cooperative.... they are claiming that she continually asks the same questions about set-up over and over again, and doesn't seem to want to learn how to do it. Winston on the other hand, could be more proactive in determining what is a business related model issue and an IT issue and ask for help from research. I think you Debbie and I need to work quite hard to get them to play nicely. I have asked Tanya and Winston to go ahead and work very closely together over the next few days....and Debbie Brackett and I will review their progress on Friday. In the meantime l'll be looking at setting up a working test environment that doesn't involve my main Quant guys in day to to day setup issues as a longer term solution. Regards Steve
|