VLAN Troubles

Brielle Bruns bruns at 2mbit.com
Tue Mar 6 23:48:44 UTC 2012


On 3/6/12 4:33 PM, Joe Greco wrote:
> I don't think the 1655 and the 5324 share ancestry.

I'm pretty sure they don't either, but never know.


>
> Dell does what lots of companies do: they outsourced.  The Dell 5_2_24
> was a catastrophic device that was based on the same hardware platform
> as the Foundry Edgeiron 24G and the SMC 8624T, 3Com's 3824 ...  the
> only difference in many cases being paint and firmware.  All of these
> were actually made by Accton, who sold it as the ES4624, and the early
> revisions had a catastrophic failure mode that would result in the two
> halves of the switch losing communications with each other, or something
> like that, hopefully I'll be forgiven for the technical handwaving, and
> eventually firmware workarounds "fixed" the switch, but (I think?)
> Foundry led the pack on that, and so you'd come across Dell gear with
> Foundry firmware or stuff like that, done by people desperate to stop
> their switches from going wonky every few weeks.

Ah yes, the EIF24G.  I have two of the -A models here and a pallet of 
them in a warehouse.

I know exactly the failures you are talking about - in our case, we got 
the switches with QC tags stating "Dead ports".  A firmware update and 
the ports magically came back and started working again.  Originally 
figured an ASIC or similar had gone south and taken out a group of ports.

I'm actually pretty happy with the switches, they're pretty durable and 
we've got the two at home acting as the 'edge' switches for things like 
the HTPC, NAS, DirecTV receivers, etc.

Before we needed the extra ports, had the foundry 10GCF which has a 
similar less then stellar history.  Mess of a witch, but it held up well 
for 2 years until the EIF24G-As came.

I believe both the 10GCF and the EIF24G's are of the same vintage of 
Accton hardware.




>
> I don't think I ever did identify the source of the 5324 fully, I think
> I concluded that it was somewhat unique to Dell.  It lacked most of the
> other quirks common to cheap switches like the Accton (broadcast domain
> issues, anyone?) and was, at the time, probably one of the best deals in
> managed switching.  It only had a few goofs that I could complain about,
> including the lack of 64-bit interface counters and the Ciscoesque-but-
> not-quite syntax.  For the most part, I've heard that their newer
> products are pretty good too, though usually there are tradeoffs.
>
> obDisclosure: We run a bunch of 5324's, and don't seem to have any
> issues with them.


I got a bit... frustrated with the switch modules on the 1655 many 
times.  CLI makes me want to puke.  Would get into a wonky state, and I 
had to factory reset it via the web ui.  Finally just gave up with the 
CLI and used the web ui to configure it.






-- 
Brielle Bruns
The Summit Open Source Development Group
http://www.sosdg.org    /     http://www.ahbl.org




More information about the NANOG mailing list