| |
N1URO > PACKET 15.01.19 17:19l 35 Lines 1584 Bytes #999 (0) @ WW
BID : 15928_N1URO
Read: GUEST
Subj: Re: UTIHZM > Node Hopping
Path: IW8PGT<IR2UBX<F1OYP<ON0AR<GB7CIP<N1URO
Sent: 190115/1604Z @:N1URO.#CCT.CT.USA.NOAM #:15928 [Unionville] $:15928_N1URO
From: N1URO@N1URO.#CCT.CT.USA.NOAM
To : PACKET@WW
Sergej (et al)
The difference between NetRom and FlexNet is that with each NetRom frame
you often see two that are sent and acknowledged - one being the ax.25
virtual circuit and the other being the NetRom circuit. The ax.25 circuit
is where a good part of your error correction lives, however with axUDP
links that use a straight stream, often that layer of error correction
gets lost in the shuffle and can either retry an excessive amount of
times causing lag or the connection may even get dropped from excessive
retries.
With Flex, there's no NetRom protocol overhead taking up at least 20 bytes
of the 256 byte MTU and the level of error correction isn't as necessary
thus making it appear to be a LOT faster.
As for the nodes in the large (what I call) INP3 CLOUD, because of how INP3
works using round-trip timers between sites, and the fact some sites may
experience lags now and again (such as a sysop streaming video) this will
cause fairly instant path changes and not everyone within the "cloud" will
sync concurrently causing ghosted paths.
While I know INP3 also carries other data within itself that may be useful
such as IP route informations, and in theory it's supposed to be a more
realtime environment it's not really suited for an internet based linked
system and often causes more harm than good as you've noticed. This is why
on my LinuXnet system I don't share it's NetRom nodes with my URONode.
I hope this helps explain a bit as to what's occurring.
73 de N1URO
---
SendBBS v1.1 by N1URO for LinFBB
Read previous mail | Read next mail
| |
