MLAG+ESRP redundancy Question

[EtherNation_Use]

Create Date: Mar 27 2012 3:05PM

Hi,


I need to interconnect 2 sites geographically separated. Both sites has the same topology as follows:

- 15 Servers aprox, each belongs to 3 or 4 VLAN's. All of those servers connected to X670-48x-FB (Stacking)
- For inter-site connection 4 Blackdiamonds 8810.

I'm planning to implant 2 tier MLAG on each site for redundancy, first for server redundancy LAB and second for inter-switch redundancy using MLAG + ESRP per VLAN (summit <-> Blackdiamond). For inter-site redundancy MLAG + ESRP too as you can see on the following diagram.


picasaweb.google.com/lh/photo/bLjNy2WVaz7nHg2kWyf9ENMTjNZETYmyPJy0liipFm0

My question is ¿Is there any limitation implementing MLAG+ESRP? ¿Can be configured in that way?


BR
JC

(from Juan Carlos_Torres)

[EtherNation_Use]

Create Date: Mar 27 2012 5:24PM

Hi,


That was exactly my concern. As ESRP has L2 and L3 functions, I was worried about ESRP layer 2 functions interferes with MLAG . So, as you mentioned if I configure on ESRP all ports belonging to MLAG as HA I can avoid that problem. Thanks a lot for the information.

BR
JC

(from Juan Carlos_Torres)

[EtherNation_Use]

Create Date: Mar 27 2012 3:43PM

Hey JC

Sorry can't see the diagram as i am responding via my phone. As you are probably aware ERSP is similar to VRRP except that it also can protect L2 redundancy as well. The other difference to ESRP vs VRRP is the protocol is L2 it doesn't require L3 in order to work. This means that it is great for L2 desgins. In your case there is already a L2 redundancy we need to tell ESRP to ignore L2. The way to do that is to mark ports as HA ports whixh means they are Host Attached ports. This tells ESRP that there are no loops on this port so always keep it open on the secondary switch. The don't count port setting tells ESRP not to count that port in its port count calculations. ESRP can used number of active ports to determine which switch will be primary. The switch with the most active ports is always master.

One thing to always remember is that both switches in ESRP need to be set up the same way no exceptions. Because it is more of a configuration to ESRP it is usually easier to use VRRP for just L3 redundancy.

Does that help?

P (from Paul_Russo)

[EtherNation_Use]

Create Date: Mar 27 2012 3:31PM

Hi,

What do you mean as "Do not count port"? . As I understand on your post, I can implement the topology detailed on the diagram (did you see it?). What I'm planning to do is configure MLAG and ESRP on same links, MLAG for Layer 2 redundancy ( on server/switches side ) and MLAG+ ESRP for layer 2/3 redundancy (all inter-switch connections).

BR
JC

(from Juan Carlos_Torres)

[EtherNation_Use]

Create Date: Mar 27 2012 3:13PM

ESRP has the same limitation as EAPS or STP. The ISC port cannot be blocked. In regards to ESRP that means that the ISC needs to be a HA port and most likeli a do not count port.

I would recommend having all of the ports used as MLAG as HA. Depending on your failover algorithm you may want the server ports as don't count.

Hope that helps.

P (from Paul_Russo)