Have you used any profiling tools to determine this, or is this just a back-of-the-envelop calculation?
While implementing only read() and not read(byte[]) obviously could stand to be improved, when I have done stress testing in conjuction with a profiling tool, I have not found this to consume a noticeable percentage of CPU. -- Serge Knystautas Loki Technologies - Unstoppable Websites http://www.lokitech.com/ Sandeep Sharma wrote: > Hi! > > I have some doubts regarding the effeciency of the > CharTerminatedInputStream : > > Processing every byte of the message will shoot up CPU > usage like any thing for example if JAMES receive a > message of 8K then there will be > > 8 * 1024 = 8192 read() function calls > > further the overhead of calling read() function 8192 > times + the processing time will consume lots of CPU > cycles consequently reducing the overall capacity of > the JAMES to handle mails > > I would advice JAMES developers to use Transaction > Cost Analysis (TRC) method to performance test JAMES > Mail Server > > Sandiep > CTO Infosoft TechNet Private Limited > http://www.infosoftin.com -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
