One of the most common questions is how much overhead do I need to account for on my transport network? I have put together a quick list to help when you are calculating your overhead.
-GRE (IP Protocol 47) (RFC 2784): 24 bytes (20 byte IPv4 header, 4 byte GRE header)
-6in4 encapsulation (IP Protocol 41, RFC 4213): 20 bytes
-4in6 encapsulation (e.g. DS-Lite RFC 6333): 40 bytes
Addition IPv4 header:20 bytes
73 bytes for ESP-AES-256 and ESP-SHA-HMAC overhead (overhead depends on transport or tunnel mode and the encryption/authentication algorithm and HMAC)
-MPLS: 4 bytes for each label in the stack
-802.1Q tag: 4 bytes
Q-in-Q: 8 bytes
-VXLAN: 50 bytes
-OTV: 42 bytes
Some rules of thumb when setting MTUs. You won’t get fragmentation if your L2 MTU is higher than your L3 MTU. This is just not the setting, but the actual overhead in use. Just setting it to a number doesn’t necessarily make it right. The above list will help you calculate the minimum MTU you may need. I try to get gear that supports a 1548 MTU and set everything to that. Makes it simple. I still want to know how much MTU I am utilizing because it helps me validate my designs.
The most important rule of thumb is you won’t get fragmentation if your l3 MTU is less than your L2 MTU.
As you get more and more into Cisco Data Center terminology you come across the term DCI. DCI is a Data Center Interconnect. DCI’s typically come in 3 categories.
Dark Fiber (CWDM/DWDM)
MPLS Layer 2 VPN (VPWS/VPLS)
MPLS Layer 3 VPN
A DCI is basically a LAN extension over one of the above methods.