N3024 inter-switch vLan routing

Question

Hi,

Apologies if this is a stupid question, but I have a client with 3 geographically separate sites who wants to have an N3024 at each site and also have them talk to each other. These switches each have an ESXi box and EQL storage box attached. They need to replicate between each other.

Each site has 4 IP ranges available...all on a 10.148.x.x

Site 1 uses 10.148.1.x and 10.148.3.x, Site 2 uses 10.148.9.x and 10.148.11.x and Site 3 uses 10.148.13.x and 10.148.15.x

The vLans in question for each site will be the third octet, so we have vLans 3, 9, 11 etc.

My question is how to get them to route the vLans to each other correctly. I have read what I can find online and I'm still confused. Most documentation seems to suggest the use of a Layer 3 device in the middle of everything, but these N3024s are already layer 3 right?

It's also suggested that each vLan have it's own IP. Which I have done...but do I need to assign the vLan IP on every switch or just the one to which the vLan is native? Example - site3 uses 10.148.3.x and so I have created and assigned that vLan an IP on that switch, but do I need to assign that vLan an IP on the other switches? If I do then I get all sorts of IP conflict messages, but if I don't then the switches are unable to ping their neighbours vLan addresses.

This is a part config I have used. I know it's going to be something simple that I am missing, but I have been looking at this for a while now and have got to the point where I can't see the wood for the trees.

ip routing
interface vlan 1
exit
interface vlan 3
exit
interface vlan 9
ip address 10.148.9.20 255.255.255.0
exit
interface vlan 11
ip address 10.148.11.20 255.255.255.0
exit
interface vlan 13
exit
interface vlan 15
exit

I have set port 1 to trunk and allowed all the vLans through, and set the rest of the ports to 'general' mode. I also tried 'access' but this didn't seem to work either.

interface Gi1/0/1
switchport mode trunk
switchport general allowed vlan add 3,9,11,13,15 tagged
switchport general allowed vlan add 1 tagged
exit
!
interface Gi1/0/2
switchport mode general
exit

...etc

So...can anyone help? Could anyone provide an exact config for how to route vLans between switches? Can I do it with only three N3024s or do I need another device in there somewhere? What is the correct settings for the ports? Should one switch act as a master switch and hold all IPs for all vLans, or should each switch hold only the vLans local to that site? And where does the default gateway fit into all this - do I need one for each vLan?

Thanks in advance for any assistance.

Duncan

Duncan_A_H · Answer

The sites are all on the same domain, though I suspect the sites connect at some point through the internet.

Everything is currently in the lab though, and right next to each other. We can connect the switches either directly or through a little 5port gig hub.

If you need any more information on the setup please do ask - it's possible I could have left out important information in the description above.

Thank you

Duncan_A_H · Answer

Thanks...I am on site next week so will be able to try it then.

I have also been advised by someone at Dell ProSupport to select one of the switches as a 'master' and define all vLans on that and assign them IPs. Then enable routing on that switch only and it should be able to route the packets around as required.

I will update as to how these strategies work, or not, later next week.

Duncan_A_H · Answer

OK...thanks for all the advice - here is an update.

I managed to get the switches to talk to each other. I defined the vLans on all switches, but only assigned them IPs on the switches to which the vLans were local.

The problem I have now is with EQL replication. Even though I can login to the EQL unit on serial and ping the partner iSCSI IP (and vice versa) the two units will not replicate.

Is iSCSi replication supported over this type of link? EQL1 iSCSi is on vLan 2 and EQL2 iSCSI is on vLan 11 - they can ping each other so surely replication should go across? When I try to set it up all I get is 'Partner is not configured'...yet all the details are (as far as I can see) correct.

Has anyone had any experience with EQL replication across different vLans?

Thanks

Duncan

Duncan_A_H · Answer

OK, thank you...sorry about the repeat posting.

Duncan_A_H · Answer

Update - vLan routing appears to be working OK, in that one switch can ping addresses on other switches OK.

The whole EQL replication issue is down to the network setup for sure, but would appear to be more with the fact that the Group IP address is not the same for all units (I had some trouble initially getting the hang of Group IP, iSCSi IP and mgmt IP).

I have tried changing the group IP of an array with the cli command >grpparams group-ipaddress x.x.x.x

...but it just said it couldn't change it as there was no netmask. But if I try to add a netmask at the end of that command it just says 'too many parameters'.

Is it possible to change the Group IP of an EqualLogic without losing all the data on it? Downtime is OK as the system is not in Production yet, but losing the array data is not.

Thank you

Duncan

Duncan_A_H · Answer

Thanks Daniel - I am pretty sure I tried that first time round and it didn't work, but I will try it again and hopefully attach a screenshot of the output.

Duncan_A_H · Answer

OK...this is all sorted now. Turns out I had a few misconceptions about things, and also was trying to make it far too complicated.

So...in case anyone else finds themselves in the situation of having to get a few Dell switches to talk to each other while at the same time getting some EQL units to replicate then maybe this will help.

Firstly, I was trying to get 3 EQL units to replicate while they all had the same Group IP. Yes, I know now...

Anyway, once these Group IPs were changed to x.x.x.1, .2 and .3 the units could all see each other.

I also got the vLan routing working. Had to define each vLan on each switch and then assign it an IP on that switch. Every vLan needed to have it's own IP...in effect, every switch had an IP for that vLan...and after that pings were able to go across the network unfettered.

I also defined static routes on the switches to define next-hop destinations, just to be sure.

Don't forget to run the 'iscsi enable' and 'ip routing' global commands on the switches too.

Thanks

Duncan

Networking General

N3024 inter-switch vLan routing

Was this post helpful?