Edit: Disregard: I was interchanging Flash and NVRAM as the same thing

I ran into an inconsistency in my study material on VTP when it comes to Client mode. Cisco Network Academy explicitly explains that when a switch is configured as a Client, Vlan information is stored in RAM, not NVRAM (Lan Switching and Wireless Chapter 4 section 4.1.1.4 & 4.1.1.5)



The CisoPress ICND2 Official Certification Guide explicitly states it IS stored in NVRAM on both Clients and Servers. (page 20-23)

I haven't labed it to check which one is it is. Maybe it's IOS or platform dependent. Though each seem to strictly use the 2960 as their base platform. I'm leaning towards the Cisco Academy of it being in RAM and not Flash/NVRAM. The Academy put a lot of emphasis on it not being in RAM with a "drop in the slot" interactive slides and even a Chapter test question directly stated at clients not having vlan database info stored in Flash, only RAM.. It is kind of odd though because the CiscoPress does also explicitly express that Clients do store their vlan databases in NVRAM and even present a "key topic".

Hummm.. So, which one is it?

I have never tested it, but the books that I have read explicitly state that VTP is not saved to NVRAM. The client sends a VTP request for the database when reloaded. My book explicitly says, "Does not save VLAN configuration in nonvolatile RAM (NVRAM)"

Which Book?

Edit: CCNP Switch book sides with CiscoPress ICND2..

CCNP SWITCH 642-813
Official Certification Guide
David Hucaby, CCIE No. 4594


Page 90



Damn-it I'm confused. How can a client have a Vlan database if it's just being powered on and not suppose to have anything stored in NVRAM..

This was straight from the CCNP SWITCH Foundation Learning Guide that I'm also reading. I also distinctly remember the passage that you quoted. How *I* interpreted it is this:

As a VTP Client, the VLAN database is not stored on a reload. However, when booting up, the client does send a summary advertisement with the revision number and if it does receive an inferior advertisement from the server, it sends a subset advertisement containing the contents of its own VLAN database (which should be empty outside of the 5 static VLANs 1, 1002-1005).

I could be wrong, but that's how I understood it. Hopefully someone can confirm or deny our thoughts.

Scott; it's one of those kinda-sorta questions. The vlan.dat file should also hold VTP information - for example, if you set a switch as a VTP client, then delete the vlan.dat file I think you'll find that the switch reverts to VTP server mode. I'll stand to be corrected on this second point, but I don't think that the vlan information itself is stored in the flash file - that should be held in NVRAM and 'lost' on rebooting the device until such time as a VTP advertisement request has been responded to or a summary/subset has been recieved.
Of course, I'm remembering this from my CCNA study so you'll have to take what I say with a pinch of salt (or lab time, if possible).
I'll try to remember to lab this up when I get some hardware for my switching lab next year.

EDIT:
http://www.cisco.com/en/US/tech/tk389/t ... 4c52.shtml
https://supportforums.cisco.com/message/596089

humm. I'm starting to come around.. It's still foggy, I'll lab it tonight. Thanks guys..

I think the only way to get an answer is to do a lab.
But since VTP clients are known to have caused VTP bombs, I find it likely that they do in fact save the information upon reboot.
But it could also be explained by reconnecting a switch that hasn't been connected for a while but powered on for the duration.

Halo's link to the CSC did have some nice information, which seems to confirm my thoughts. But I'm a bit unsure if he states that the VTP client still had vlans in it after a reboot or just the VTP config. Either way its still just as dangerous to connect a VTP client as a VTP server to a network.

Yeah, I'm going to lab it tonight.. I get it, but I don't. What's in the Flash:Vlan.dat and what's in the NVRAM is getting me twisted.

Edit:

Wow, I'm sitting here reading my thoughts and I just want to get up and walk away from the keyboard! I've been thinking of Flash (non volatile) as the same as NVRAM. Nice one Scott, keep up the good work..

I guess the obvious lesson here is to clear the revision number on any new switch running VTP before connecting it to your network by either changing the VTP domain to an unused domain and back to the preferred domain or by setting the VTP mode to transparent and then back to server/client.

No, I'm pretty sure the obvious lesson here is to know what the hell you're talking about before staring a thread. But that too.

If nothing else, you made me re-potentiate those neural pathways I'd set up to get my head around VTP back when I studied for my CCNA. Heaven knows I'll need 'em again when I get the SWITCH FLG and OCG on my birthday. Which is this friday. Just sayin'.
*cough*someonebuymeaswitch*cough*

im just finished reading this vtp section and am gonna have to play with some stuff til it sits right. i havent been able to go through the PT labs yet cos im at work and left my portable PT app at home . also college is shut til next year so no chance of playing around on kit either .
its sitting ok but i need to see it in action, implement, troubleshoot etc more so i get that "aahhhh" feeling.

Its in the Vlan.Dat file...I am reading up a little bit / reviewing for my switch exam... but from what I remember with VTP Bombs where I work was to delete the VLAN.dat data base / turn the Switch onto transparent mode while putting it on the network.... but you are correct to where it will NOT store it into NVRAM.

_________________
"I will prepare and some day my chance will come." - Abraham Lincoln
http://danielhertzberg.wordpress.com - I blog about networks!