[ENet-discuss] ENet scalability benchmark data (Lee Salzman)

[Espen Overaae]

On 3/6/08, Ruud van Gaal <ruud at racer.nl> wrote:
> ...
>  > Note that those sorts of benchmarks are somewhat artificial,
>  > in that they go against the grain of ENet's design.
>
>  I also wondered what this benchmark for, other than for the benchmark
>  itself. ;-)
>  ENet looks like a gaming network library, where you have to serve at most
>  about 50 persons in a regular client-server or peer-peer network setup.
>  Situations where you'd get 1000+ connections: a webserver (making your own
>  Apache) and Massive Multiplayer Online Games. Both of which I think are
>  certainly not the target of ENet.

They certainly are not, but there really aren't many alternatives when
you want to combine low latency, reliable relivery, and thousands of
clients.

If it makes you feel any better, I performed a different test where I
connected 256 clients and sent a lot more packets from each, and
upwards of 80 k packets managed to get through to the server each
second. That's several times more than RakNet and ICE claim.


Espen Overaae