VPC+, aka L3 on back-to-back vPC domains

Question

Hi,
Please consider this scenario, where L2 VLANS are spanning 2 data centers and where R1-R4 are L2/L3 N7K routers (replacing existing 6K).
(I wish VSS would be available also in N7K to make life 10x easier!!).
                      R1                  |                     R2
                       ||                    |                     ||
vPC peer-link  ||    =======MAN=======   ||  vPC peer-link
                       ||                    |                     ||
                      R3                  |                    R4
                               Site A         Site B
Attached to R1 and R3 there are (dual-attached via 6K access switches) servers that may need to communicate to other servers in the same VLAN on the other side of the MAN. Over the MAN the VLANs are trunked, so its fine. This traffic can go over R1 or R3 both for L2 (vPC) and for L3 (HSRP vPC enhancements).
Anway, there is also a global OSPF domain for inter-VLANs communication and for going outside the DCs via other routers attached to the above cloud.
/* Style Definitions */
table.MsoNormalTable
{mso-style-name:"Table Normal";
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-qformat:yes;
mso-style-parent:"";
mso-padding-alt:0cm 5.4pt 0cm 5.4pt;
mso-para-margin:0cm;
mso-para-margin-bottom:.0001pt;
mso-pagination:widow-orphan;
font-size:10.0pt;
font-family:"Times New Roman","serif";}
I've heard there is a kind of enhancement request (or bug?,
/* Style Definitions */
table.MsoNormalTable
{mso-style-name:"Table Normal";
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-qformat:yes;
mso-style-parent:"";
mso-padding-alt:0cm 5.4pt 0cm 5.4pt;
mso-para-margin:0cm;
mso-para-margin-bottom:.0001pt;
mso-pagination:widow-orphan;
font-size:10.0pt;
font-family:"Times New Roman","serif";}
CSCtc71813) to have this kind of back-to-back vPC scenario to handle transparently L3 data (peer-gateway command should deliver also control-plane L3-info??). There are 2 workarounds available for this design:
1.       Define an additional router-in-a-stick using an extra VDC on each 7k. In this case, for example for R1 we would use 3 VDCs: 1 VDC for admin, 1 VDC for L2, 1 VDC for R1.
2.       Define static routes to tell each 7k how to reach the other 7k L3 next-hops.
a) What is the best workaround to choose in order to smooth the upgrade later to the version of vPC that will handles this issue?
b) Are there any more caveats I dont see? I havent seeen any link in CCO, so I am unsure how to proceed the design.
c) I would be tempted to think that using additional static routes is a better choice because it would easier to remove them once vPC+ is there.
What static routes shall I add? R1 to R2, R1 to R4 and so on and so forth? I miss the details of this implementation.
d) How would vPC+ looks like once (when?) is there?
Thansk for your valuable input in advance.
G.

Answer 1

To expand on Lucian's comment, because I'm sure the next though will be...can I run OSFP over a vlan and just carry THAT over my VPC.  You don't want to do this either.
We don't support running routing protocols over VPC enabled VLANs.
What happens is that your 6500 will form routing adj with each Nexus 7000....lets say Nexus 7000-1 and 7000-2.  Note my picture below.
Lets say that R1 is trying to send to a network that is behind R2.  R1 is adj to 7000-1 and 7000-2...we have equal cost paths.  CEF chooses that 7000-1 to route the packet, however Etherchannel load balancing chooses the physical link to 7000-2.  7000-2 will need to switch the packet over the VPC peer-link to 7000-1.  7000-1 receives the packet and tries to send it out VPC member port to R2....however egress port drops the packet.  This happens because we don't allow packets received from VPC member link send over VPC peer-link to be sent out another VPC member link.
I'd suggest to run an L3 link from your 6500 to each Nexus 7000 if you do want to do L3 on it.