T
Thomas Hansen
Howdy, I've been working on this solution for quite some time where we
basically turn the HTTP protocol upside down and manages to have a
"Message Que" on the server from which we can post "events" back to the
client upon.
Basically it works by having a VERY LONG timeout on an Ajax HTTP
Request that basically never returns before something happens on the
server that raises an event, and when an event is raised that event is
returned to the client on that "locked" http Ajax request.
For those interested in reading about it have a look at my blog her:
http://www.frostinnovation.com/Blog.aspx
And for those interested in seeing a sample implementation of it have a
look here:
http://www.frostinnovation.com/PresentasjonUtviklerForum/Chat.aspx
Then to my question, after reviewing the basic concept, I've managed to
pinpoint a couple of problem areas, but I'm not really sure if I've
managed to pinpoint them all?
My biggest concern is the THREADS that's locked on the server...
How many threads can e.g. Windows Server 2003 have in the NETWORKING
process before it "crashes" completely...?
I think the general number for ANY process on Windows Server 2003 is
something about 32000 or something...?
Is there any other reason as to what might happen when this solution is
scaled extremely upwards?
Eager to get responses and discussions around this theme, regards
PS!
For those interested in using this for themself, we're interested in
getting in contact with early adopters and beta testers for a framework
we're developing called Gaia WOS which includes among other things this
concept.
Thomas Hansen
basically turn the HTTP protocol upside down and manages to have a
"Message Que" on the server from which we can post "events" back to the
client upon.
Basically it works by having a VERY LONG timeout on an Ajax HTTP
Request that basically never returns before something happens on the
server that raises an event, and when an event is raised that event is
returned to the client on that "locked" http Ajax request.
For those interested in reading about it have a look at my blog her:
http://www.frostinnovation.com/Blog.aspx
And for those interested in seeing a sample implementation of it have a
look here:
http://www.frostinnovation.com/PresentasjonUtviklerForum/Chat.aspx
Then to my question, after reviewing the basic concept, I've managed to
pinpoint a couple of problem areas, but I'm not really sure if I've
managed to pinpoint them all?
My biggest concern is the THREADS that's locked on the server...
How many threads can e.g. Windows Server 2003 have in the NETWORKING
process before it "crashes" completely...?
I think the general number for ANY process on Windows Server 2003 is
something about 32000 or something...?
Is there any other reason as to what might happen when this solution is
scaled extremely upwards?
Eager to get responses and discussions around this theme, regards
PS!
For those interested in using this for themself, we're interested in
getting in contact with early adopters and beta testers for a framework
we're developing called Gaia WOS which includes among other things this
concept.
Thomas Hansen