Re: Anyone working on better transaction locking?
От | Jan Wieck |
---|---|
Тема | Re: Anyone working on better transaction locking? |
Дата | |
Msg-id | 3E997F67.E07B1C4F@Yahoo.com обсуждение исходный текст |
Ответ на | Re: Anyone working on better transaction locking? (Jan Wieck <JanWieck@Yahoo.com>) |
Список | pgsql-hackers |
Ron Peacetree wrote: > > "Jan Wieck" <JanWieck@Yahoo.com> wrote in message > > And you are comparing what? Just pure features and/or > > performance, or total cost of ownership for your > > particular case? > > > Technical Analysis and Business Analysis are two separate, and equally > necessary, activities. However, before one can accurately measure > things like Total Cost of Ownership, one needs to have accurately and > sufficiently characterized what will be owned and one's choices as to > what could be owned... Okay, so you are doing the technical analysis for now. > [...] > However, a correctly done Technical Analysis =should= be reasonably > portable since among other things you don't want to have to start all > over if your company's business or business model changes. Clearly > Business Analysis is very context dependant. However, doing a technical analysis correctly does not mean to blindly ask about all the advanced features for each subsystem. The technical analysis is part of the entire evaluation process. That process starts with collecting the business requirements and continues with specifying the technical requirements based on that. Not the other way round, because technology should not drive, it should serve (unless the technology in question is your business). Possible changes in business model might slightly change the technical requirements in the future, so an appropriate security margin is added. But the attempt to build canned technical analysis for later reuse is what leads to the worst solutions. How good is a decision based on 2 year old technical information? Now all the possible candidates get compared against these "requirements". That candidate "O" has the super duper buzzword feature "XYZ" candidate "P" does not have is of very little importance unless "XYZ" is somewhere on the requirements list. The availability of that extra feature will not result in any gain here. In an earlier eMail you pointed out that 2 phase commit is essential for SMP/distributed applications. I know well what a distributed application is, but what in the world is an SMP application? Jan -- #======================================================================# # It's easier to get forgiveness for being wrong than for being right. # # Let's break this rule - forgive me. # #================================================== JanWieck@Yahoo.com #
В списке pgsql-hackers по дате отправления: