Just in a joking "friends don't let friends do x" sort of way, and even that was more directed at the original author.

You're right about RSTP, but the daisy chaining hosts thing is just bad mojo in general.  Some random host between to others conversion gets wonky, you have to figure out which is causing a problem, and why.  Enterprise networking doesn't even like to daisy chain switches anymore, let alone hosts.

Not to mention it's entirely unsupportable by anyone but the original author setting up hosts as bridges, as no one else would do something like this in the business world, let alone even at home.

Just go with a switch and keep your life easy.

-mb


On Fri, Oct 4, 2019 at 11:57 PM Stephen Partington <cryptworks@gmail.com> wrote:
It appears that some buttons were pushed. My initial reading did suggest rstp was very good to maintain switch to switch redundancy, which is what i initially thought. Re reading your initial email I am still very curious about why you were looking to use rstp as a nic to nic design. 

On Fri, Oct 4, 2019, 11:48 PM Michael Butash <michael@butash.net> wrote:
I really don't get any anyone in their right mind would do this other than an experiment to say they can/did.  Host ethernet chaining is not what (Rapid) Spanning Tree Protocol was designed for, and with modern (or old) switching, there is no reason to.  As a network engineer for 20 years, it offends certain sensibilities as something you should never do.

There is a reason people have been using ethernet hubs/switches for 30 years now - speed and simplicity.  If you walked into any sort of enterprise or business with any network knowledge and proposed that, someone might just fire you.

Switches are designed to forward quickly and effectively, some as low as 350 down to 8 nano seconds these days with special nics,  Even a server cpu bridging at a kernel level will *never* do so as quickly as that, particularly cumulative latency in a chain.  Servers that do have more than one nic certainly aren't intended to be daisy-chained, rather they home each nic to multiple vlan segments, or they aggregate nics as active/passive or active/active link aggregation to multiple switches (redundancy).  Hosts as a rule should NEVER talk spanning-tree, only switch to switch.

Just... don't ever chain hosts like that, particularly not if said client is paying you for a network solution.  Get a switch or multiple with as many ports as you need.  Ebay is always good for slightly older kit, and just get a spare to keep around just in case.

If you're *that* interested in networking to build that sort of science experiment, pick up a CCNA switching book to learn why you're barking up the wrong tree.

-mb


On Fri, Oct 4, 2019 at 6:34 PM Stephen Partington <cryptworks@gmail.com> wrote:
I am still wrapping my head around why this was the root design.

I am not sure what gains you have vs having a pair of switches for redundancy. time to research RSTP.

On Fri, Oct 4, 2019 at 3:34 PM kelly stephenson <stephenson2773@gmail.com> wrote:
Looking for some networking advice from the group.

The system I have has several devices connected in a ring configuration using one Ethernet port IN and one Ethernet port out.  The system uses RSTP for loop free operation.  The idea is simplicity for installation, you just unplug and plugin a new device in the ring plus you gain redundancy, if one Ethernet cable breaks you still have another one.  This works but my client has never had more then a half dozen devices on the network yet.
When I say devices just imagine very large machines.  The number of devices could be as many as 100 in the ring or network.  Everything I've researched on RSTP says over 8 devices and its not effective/efficient so I'm researching other Ethernet failover/failsafe/redundant solutions.
So, the local network configuration needs to scale up to 100 devices, have redundancy, and low latency for M2M control.  Any thoughts?  

Thanks
Kelly
---------------------------------------------------
PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
To subscribe, unsubscribe, or to change your mail settings:
https://lists.phxlinux.org/mailman/listinfo/plug-discuss


--
A mouse trap, placed on top of your alarm clock, will prevent you from rolling over and going back to sleep after you hit the snooze button.

Stephen

---------------------------------------------------
PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
To subscribe, unsubscribe, or to change your mail settings:
https://lists.phxlinux.org/mailman/listinfo/plug-discuss
---------------------------------------------------
PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
To subscribe, unsubscribe, or to change your mail settings:
https://lists.phxlinux.org/mailman/listinfo/plug-discuss
---------------------------------------------------
PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
To subscribe, unsubscribe, or to change your mail settings:
https://lists.phxlinux.org/mailman/listinfo/plug-discuss