Re:.NET Core WebSockets – Cross-Platform WebSockets From DMZ...
by Quarrelsome on Thursday January 05, 2017 @05:05AM
(#13326003)
I'm somewhat new to websockets but that code looks suspiciously like the "non-scaling" broadcast mechanisms that .NET remoting used to offer. Can I safely assume this mechanism for pushing updates to clients doesn't scale in the same way most use cases (except single-use client requests) in .NET remoting don't scale too?
[ Parent
| on HN ]
Re:.NET Core WebSockets – Cross-Platform WebSockets From DMZ...
by statictype on Thursday January 05, 2017 @05:23AM
(#13326066)
Could you explain the scaling issues with .NET remoting
I knew there were issues but didn't know the specifics.
Re:.NET Core WebSockets – Cross-Platform WebSockets From DMZ...
by Quarrelsome on Thursday January 05, 2017 @05:36AM
(#13326111)
silver bullet. The idea was a wonderful new world where you could have remoted objects and not have to care about where and how they were.
Except, you did have to care. The process of not caring resulted in a world where round-trips were not isolated and an excess of them occurred resulting in awful performance.
I forget the precise table but there were many means of using .NET remoting and basically all the "fun stuff": client-activated server-based objects, sever-based publishing (i.e. like this implementation of websockets), and server-based singletons all had immense problems with scaling. The only means of using .NET remoting that did scale was one-time-use server based services so to scale you'd end up just using .NET remoting like webservices/RPCs or REST APIs and effectively 3/4s of the features were "worthless".
It was an interesting diversion and was eventually "replaced" by WCF. It is still very much used under the covers of a lot of things though. IIRC COM still uses .NET remoting or aspects of it to communicate with .NET apps.