Thanks Arnie

You may be right, and certainly ARE right about sailboats being slow. Actually, I don't care about latency much at all. In fact, we will delay/obfuscate the data by as much as four hours anyway in order to reduce tactical use by competitors. It's really the issue of an AJAX request staying open on a conventional server apparently ties up resources.

Your suggestion of simply sticking with conventional polling, but reducing the cycle count, is an excellent one, and a heckuva lot easier to implement.

I had stumbled across websockets at the same link you provided just a moment ago. I guess the issue there is it is not fully implemented or deployed, but WILL BE.

I'M STILL interested in other views on this. With 90k visitors a day, reducing server load while delivering an exciting experience is a good thing.

Michael




On 5/20/2011 7:52 AM, Arnie Shore wrote:
I'll certainly defer to others who've done measurements here, but from some experience with our open source computer-aided-dispatch application:

WRT " ... It seems that the "long polling" method or, worse, looped queries, can lead to server overload ... ." I expect here that the culprit is in implementers trying to minimize latency by using an excessively short cycle period, which results in hammering the server.

Riding to rescue is WebSockets - http://en.wikipedia.org/wiki/WebSockets - which supports server push, and for which many of us wait in eager anticipation. It's a nice solution to the general problem of http in a real-time environment.

At this writing, the protocol spec is fairly far along but implementation is uneven. I'll guess that it might be widely available in the latest browsers later this year.

Pending that availability, consider conventional solutions like those you've named, but with a reasonable cycle time. And live with the latency. (AS fast as these boats are, they remain sailboats!)

AS


On 5/20/2011 10:14 AM, Michael wrote:
> Greetings Mappers!
>
> I've built a trans-pacific charting application intended to be used for
> near real-time sailboat tracking.
_______________________________________________
Users mailing list
[email protected]
http://lists.osgeo.org/mailman/listinfo/openlayers-users
_______________________________________________
Users mailing list
[email protected]
http://lists.osgeo.org/mailman/listinfo/openlayers-users

Reply via email to