Logical top-of-rack (TOR) switches

A method comprising:

• executing, by a network device, a master pod comprising one or more containers and a plurality of worker pods comprising one or more containers, wherein each of the plurality of worker pods is configured to execute a corresponding a plurality of network Operating System (OS) instance of a plurality of network OS instances;
• assigning, by the master pod, each of the plurality of worker pods to a different, corresponding tenant of a plurality of tenants of a datacenter;
• allocating, by the master pod, a different subset of a plurality of interfaces of the network device to each worker pod of the plurality of worker pods;
• and
• managing, by each network OS instance of the plurality of network OS instances executed by the network device, forwarding of network traffic associated with the corresponding tenant and received via the corresponding subset of the plurality of interfaces allocated to the worker pod of the plurality of worker pods that is configured to execute the network OS instance.

Non-transitory, computer-readable media comprising instructions executable to cause processing circuitry of a network device to:

• execute a master pod comprising one or more containers and a plurality of worker pods comprising one or more containers, wherein each of the plurality of worker pods is configured to execute a corresponding network Operating System (OS) instance of a plurality of network OS instances,
• wherein the master pod is configured to assign each of the plurality of worker pods to a different, corresponding tenant of a plurality of tenants of a datacenter,
• wherein the master pod is configured to allocate a different subset of the plurality of interfaces to each worker pod of the plurality of worker pods, and
• wherein each network OS instance of the plurality of network OS instances is configured to manage forwarding of network traffic associated with the corresponding tenant and received via the corresponding subset of the plurality of interfaces allocated to the worker pod of the plurality of worker pods that is configured to execute the network OS instance.