High availability and I/O aggregation for server environments
View Patent ↗Methods and apparatus are provided for virtualizing port adapter resources such as network interface cards (NICs) used to connect servers to packet based networks. Resources are offloaded from individual servers onto a resource virtualization switch. Servers connected to the resource virtualization switch using an I/O bus connection share access to NICs. Redundancy can be provided using multipathing mechanisms implemented at individual servers or high availability mechanisms implemented at the resource virtualization switch. Switchover can occur between ports on the same port adapter, between ports on separate adapters, or between ports on separate resource virtualization switches.
1. A resource virtualization switch, comprising:
an I/O bus switch connected to a plurality of external servers through a plurality of I/O bus ports, each of the plurality of external servers comprising a separate memory address space, the plurality of external servers including a first server running a first application and a second server running a second application;
a plurality of network interface cards (NICs) connected to a PCI-Express bus switch;
a virtual network chip of a server platform connected to the plurality of I/O bus ports of the I/O bus switch and the plurality of NICs, wherein the plurality of NICs are accessible to the plurality of external servers as virtual NICs (VNICs) through the virtual network chip of the server platform, wherein a plurality of VNICs are assigned to the first application to allow for path redundancy in the event a particular NIC fails;
a processor subsystem configured to initialize an internet protocol (IP) network connection through a first NIC regardless of whether any of the plurality of external servers are connected to the resource virtualization switch.
2. The resource virtualization switch of claim 1 , wherein the first application accesses a particular VNIC as though the first application is accessing a particular NIC included in the first server.
3. The resource virtualization switch of claim 1 , wherein the first application is operable to access the plurality of VNICs for load sharing.
4. The resource virtualization switch of claim 1 , further includes a plurality of host bus adapters (HBAs).
5. The resource virtualization switch of claim 4 , wherein the plurality of HBAs are accessible to the plurality of external servers as virtual HBAs (VHBAs).
6. The resource virtualization switch of claim 5 , wherein the second application accesses a particular VHBA as though the second application is accessing a particular HBA included in the second server.
7. The resource virtualization switch of claim 4 , wherein the first server includes a first virtual NIC driver and a second virtual NIC driver, wherein the first and second virtual NIC drivers are operable to access a plurality of NICs for redundancy and load sharing.
8. The resource virtualization switch of claim 7 , wherein the first and second virtual NIC drivers support the creation of interface groups used for multipathing.
9. The resource virtualization switch of claim 1 , wherein the virtual network chip comprises a buffer manager, wherein the buffer manager manages data in a memory associated with the virtual network chip.
10. The resource virtualization switch of claim 9 , wherein the virtual network chip further comprises a queue manager, wherein the queue manager manages descriptors for the data including a reference to a memory location for the data, a length of the data, a source port for the data, and a multicast count of the data.
11. The resource virtualization switch of claim 10 , wherein the virtual network chip further comprises classifier logic, wherein the classifier logic identifies information including a packet destination server and priority for incoming data.