Wednesday, November 24, 2021

Migration of workloads in a DC environment. Is this really a problem?

Hello,

I am learning EVPN/VXLAN technology from scratch. I have figured out the basic concepts of how these protocols work. I understand that L2 traffic is transmitted over L3 underlay between VTEPs, that VXLAN = data plane and EVPN = control plane, understand how they work together, etc. But there is one thing whose value I cannot understand.

One of the features explicitly mentioned in almost every EVPN/VXLAN book is the ability to "migrate workloads (BMS, VM, container)". But I can't figure out what the actual pluses are compared to the legacy VLAN technology. Ok it's done via control plane, not via data plane. I undestand it. But what are the pros of this for say DevOps or network engineers in terms of operations tasks?

Let's say if we migrate a VM from one physical server to another which is connected to legacy network (without using EVPN/VXLAN), the new mac address location will simply be updated through a normal "flood and learn" process = the VM will send traffic, mac tables will be updated, etc. Give your opinion, please, maybe I just don't understand the problems faced by DevOps and system administrators in large DCs.



No comments:

Post a Comment