Increasing internet protocol version 6 host table scalability in top of rack switches for data center deployments

摘要

Techniques are provided for optimizing storage of address information in switch device databases. A control packet is received from a source host. An address associated with a host is determined. The address comprises a first and second address component. The first address component is stored in a first database and is mapped to an index value that is stored in the first database. The index value is also stored in a second database along with the second address component. The first database may also be examined to determine whether the first address component is stored in the first database. If the first address component is stored in the first database, the index value mapped to the first address component is retrieved. The second database is examined to determine whether the index value and the second address component are stored in the second database.

主权项

1. A method comprising:
at a switch device in a network: receiving a control packet from a source host in the network;
determining from the control packet an Internet Protocol (IP) address associated with a host in the network, wherein the IP address is divided into a first IP address component that is a subnet associated with the host and a second IP address component that is an interface identifier of the host;storing the first IP address component in a first database that is a longest prefix match (LPM) database of the switch device;mapping the first IP address component to an index value that together with the second IP address represents the host IP address and storing the index value in the first database;storing the index value and the second IP address component in a second database that is a host routing table (HRT) of the switch device; receiving a data packet including an IP address including a target subnet associated with a target host and a target interface identifier of the target host; and searching the LPM database for the target subnet and, upon finding the target subnet: retrieving from the LPM database an index value mapped to the target subnet; and searching the HRT for the index value and the target interface identifier that together represent the target host IP address; and upon finding the index value and the target interface identifier in the HRT: determining a next-hop adjacency address for the data packet; and sending the data packet to a device associated with the next-hop adjacency address.