From: Noveck, Dave (Dave.Noveck@netapp.com)
Date: 03/16/01-04:16:23 PM Z
Message-ID: <8C610D86AF6CD4119C9800B0D0499E331A6F40@red.nane.netapp.com> From: "Noveck, Dave" <Dave.Noveck@netapp.com> Subject: RE: append writes (con't) Date: Fri, 16 Mar 2001 14:16:23 -0800 Spencer Shepler wrote: > On Fri, Noveck, Dave wrote: > > > > > On Fri, Noveck, Dave wrote: > > > > > > The question is whether multiple processes appending to a log file > > > > belongs in that class. I don't think it does. That is why people > > > > keep trying to solve it. The problem isn't that it is an > > > > > > "people" refers to customers or system vendors? > > > > In this case, "people" refers to engineers, who work for system > > vendors. System vendors are usually corporations and not people, > > although I'm sure there are exceptions. > > So there must be a customer somewhere that wants this if engineers are > spending their time on this difficult problem. It would be nice to > know about the customers and what they are attempting to accomplish. People write log files for many reasons. All OS's that I know of have an append mode for this very purpose. It would be very inconvenient for each process to have it's own log due to the volume of files and the fact that you would have to post-process to get any coherent view. This is why OS's typically allow multiple processes to write, in append mode, to the same file with a coherent result, on local file systems. What are the customers that use this feature trying to accomplish? Just get their work done without wasting time doing something complex. They prefer to have a facility that makes their life simple. In a networked environment the same hold true. To have to log application data based on the machine on which the process executed is undesirable. You can program around this and some people have but it is clearly not desirable and anybody who has to kludge around this would certainly choose to spend his time otherwise.
This archive was generated by hypermail 2.1.2 : 03/04/05-01:48:41 AM Z CST