Gigabit Intel Stack Update
Added 2024-07-08 00:47:52 +0000 UTCI got the gigabit fiber uplink to the Intel Express stack working... with a surprising combo of hardware and software changes.
Comments
Sometimes. This time I didn't, I'll remember that gbic enable part (even if I forget the syntax). But for more complex configs I'll save things, for example Cisco IOS configs for things like the dial-up routers.
clabretro
2024-07-09 16:34:27 +0000 UTCDo you document all of the stuff you try, like all the steps? If so it would be awesome to see a video of what you use and how you organize it. Seems to be a step I always miss and kick myself for when I need to do the same thing again in the future.
ConnerWithAnE
2024-07-09 16:09:31 +0000 UTCI was wondering too, it'd make sense
clabretro
2024-07-09 03:21:31 +0000 UTCha! I thought about running a fresh run from the rack but I think I'll steal an unneeded one from the unfinished area I film in
clabretro
2024-07-09 03:21:07 +0000 UTCCAT6 in the gym? Never know when you will need to hardwire those new exercise machines!
Kyle Bee
2024-07-09 01:53:36 +0000 UTCI forgot to add that for most things, speed (rate of transfer) is measured in bits per second. SCSI is the notable exception to this rule. Capacity / size tends to be measured in bytes. At least historically.
DrScriptt
2024-07-08 23:59:20 +0000 UTCI wonder if the weird cable ports are a form of a Media Independent Interface (MII). https://en.wikipedia.org/wiki/Media-independent_interface There is an evolution of separating the PHY (physical) and the MAC (logic using the PHY) in equipment. This is represented in AUI, MII, , and CX4 for 10GBase-T.
DrScriptt
2024-07-08 23:22:07 +0000 UTCCheck the fiber transceivers closely. You will need to have the same mode; single vs multiple, on both ends. You should have a corresponding fiber patch cable of the same mode. There is a little bit of wiggle room in that in some situations, the wrong mode patch cord will work for short distances.
DrScriptt
2024-07-08 23:10:18 +0000 UTCThat's a great idea about 10.0.0.0, I've been thinking about exactly that and getting a full lab network so I'm not hunting and looking for available 192.168.1.* IPs on the main network, it's getting old haha.
clabretro
2024-07-08 21:23:01 +0000 UTCSure thing. In the real world bridge groups tend to only see specialty use cases, but in the lab it's really nice to know the capability exists if you need to hack something together. Re: the plan to NAT through the 7200: rather than NAT, you might consider putting the entire clab universe in 10.0.0.0/8 and static routing it from your home router to the 7200. That way you can leave the NAT on your main home router and have contiguous two-way reachability within your house without a second NAT in the middle. It'll also make it really easy to distinguish lab-sourced traffic from home-sourced traffic in a packet capture.
Eric
2024-07-08 19:25:24 +0000 UTCGreat info, thank you! The whole subinterface notion was totally new to me. I'd never heard of bridge groups either, I'll have to play around with those. When I figure out a performance testing setup my basic plan is to just NAT with the 7200 (though, I really don't even need to get it onto the internet at all -- but I like to do that in all these experiments to learn more ha).
clabretro
2024-07-08 13:47:21 +0000 UTCWild that you needed an extra stack cable for the Intel side. A lot of times the labeling on the front of enterprise network gear looks completely baffling until you read the documentation. The label around the EN light on the 1000BASE-T interface (on the Cisco side) and the rectangle around *both* a stack interface *and* the 1000BASE-SX SC interface (on the Intel side) really proved to be big clues in retrospect. I would have tripped over both of those and it's always fun watching the troubleshooting saga play out live. Errata that may prove to be helpful since you mentioned subinterfaces: There's a fun terminology distinction that was once the "how many angels can dance on the head of a pin" for network engineers: what do "router", "bridge", "switch", and "layer-3 switch" even mean? For some meanings of "router" the 7200 certainly is one. And as such, you're traditionally terminating 802.1Q demarcated segments coming from a big pile of less expensive "bridges" or "switches" onto one very expensive "router" port. In that circumstance, the "switching" or "bridging" logic ends at the router port and the concept to reach for there is, as you discovered, the subinterface. For a "layer-3 switch" the device both acts as a "switch" and a "router" and instead of subinterfaces you terminate the segment on a switch virtual interface (SVI) (`interface VlanXXX` in IOS speak). Your old Catalyst stuff probably prefers to be configured that way. Know also that if you really need to "bridge" through your 7200, IOS does have something called bridge groups. The traditional subinterface you discovered is a layer-3 subinterface: that is, it's a logical router port that exists as a child of the physical port and is distinguished by the 802.1Q tag. If instead you add the subinterface to a bridge group it will operate as a layer-2 subinterface. Then it will do the usual "switched" Ethernet flood and learn between other interfaces in the group. I'm pretty sure you can mix L2 and L3 subinterfaces, too.
Eric
2024-07-08 04:49:15 +0000 UTCIf I had to guess, the module is acting like a media converter, which is why the second stack cable is important. Weird design decision but very interesting
Joe Longendyke
2024-07-08 01:46:27 +0000 UTC
