Damn
asked on
architecture: applets over http too slow
I have a an java application accessed by huge number of applet based clients over proprietary http based protocol. Clients access a central server through various levels of proxies. A group of clients clients and last level proxy are in the same Lan (same institution's branch). Server is connected to legacy backend.
Application suffers a latency, response time problem. I think it is in part due to not that much efficient, web-service like, request based protocol. Is there any way to make this more efficient, maybe pool these connections in some way?
Distributing application is at the moment out of question.
Application suffers a latency, response time problem. I think it is in part due to not that much efficient, web-service like, request based protocol. Is there any way to make this more efficient, maybe pool these connections in some way?
Distributing application is at the moment out of question.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
ok, I guess this is really complicated. Increased to 500 points.
Well, I guess that pooling - if implemented at a proxy behind which the clients are hidden, could be a good idea - if you have means of establishing that. Let's not change the protocol right now - we'll see about that later in case pooling still doesn't help. However, as of now, I don't see anything wrong with HTTP.
Split: mayankeagle, rama_krishna580
ASKER
mayankeagle,
Well, pooling is first idea that came to my mind. My idea is having something of P2P pooling, so one client serves as proxy for a certain number of close-by applets, behind same proxy.
Of course, this only has sence if something is gained, and rerouting http requests in this way would have no impact. That's why I am considering also switching to some more efficient protocol, but I am not shure just how much pure tcp is more efficient compared to http?. And of course this could imply huge number of network restriction (firewalls) etc. problems.