This discussion is just starting to stray into 'advanced-java'
territory, indeed we had a discussion about large numbers of threads
in a single VM a month or so ago. David Chase of Natural Bridge
cited their product, as proof by example that it is possible
to support large numbers of threads on a single VM, by multiplexing
multiple java threads onto an OS thread.
However Gosling's origional design for Java threads was explicitly
that they could map 1-to-1 with the OS's threads, so for the while
we'll have to live with OS threading limitations.

As far as async servlets are concerned, I don't see a future in
the 'async get()' pattern, where the current get() method is split
into two methods, one that starts a thread to process the request,
and a second 'call-back' method that is invoked when the thread
completes in order to send the actual data.

What this discussion has raised (for me) is the issue of background
threads in the servlet engine. As I understand it, these are frowned
upon, but there are cases where they are _very_ useful, for example
use them to poll snmp variables, which ensures that we can always pop
up a historical graph of (say) bandwidth whenever a browser request
comes in.

We'd like to see some support for this in the servlet container,
with threads started at startup an (possibly) serialized at shutdown.

Tim

___________________________________________________________________________
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff SERVLET-INTEREST".

Archives: http://archives.java.sun.com/archives/servlet-interest.html
Resources: http://java.sun.com/products/servlet/external-resources.html
LISTSERV Help: http://www.lsoft.com/manuals/user/user.html

Reply via email to