Re: Re: More xpl communication problems

[Kevin Hawkins <lists@xxxxxxxxxxxxxxxxx>]

Ahh - hadn't thought of the possibility of a single hub Tony. Redundancy
is as you say the key problem with this -  also another downside is that
the network traffic would expand considerably.  Packets routed via the
loopback interface don't appear on the main Ethernet LAN so the extra
relay traffic is contained within the PC. If there was a central hub
then every receiving application (not using the 3865 port) would need
all xPL traffic relayed to it over the main LAN so the network is loaded
by the number of such xPL hub applications times the total xPL traffic ,
(albeit on different ports).

K


Mal Lansell wrote:

>Actually that was what I was getting at ;-)  Redundancy, of course
>(slaps forehead).  Turn off one PC, and the rest can keep functioning
>because they have their own hubs.
>
>That was a good explanation from Kevin though - perhaps it could go
>in the docs in place of the couple of lines that are there at the
>moment.  If the hub does indeed send to 127.0.0.1, that would also
>explain why binding my receiving socket to ADDR_ANY rather than the
>real IP worked.
>
>Mal
>
>
>--- In ukha_xpl@xxxxxxx, "Tony Tofts" <tony@x> wrote:
>
>
>>Thanks for the description Kevin, nicely explained :-)
>>
>>I suspect mal might also be hinting at having a single hub for all
>>
>>
>machines,
>
>
>>which although technically can be done wasn't since it reduces the
>>redundancy of the system.
>>
>>Regards
>>Tony
>>
>>
>
>
>