Server architecture having dedicated compute resources for processing infrastructure-related workloads

A method, comprising:

• hosting an application on a server node, wherein the server node comprises:
• a plurality of processors comprising a first processor and a second processor;
• a first input/output (I/O) processor comprising an I/O hardware interface coupled to a front-end communications network and a backend data storage network, wherein the I/O hardware interface of the first I/O processor implements converged network adapter functions to provide a data entry point and data exit point for each of the front-end communications network and the backend data storage network;
• a second I/O processor;
• a first shared memory coupled to the plurality of processors;
• a second shared memory coupled to the first and second I/O processors;
• and
• a global memory coupled to the first shared memory and the second shared memory, wherein the global memory is configured to enable communication between the first and second I/O processors and the plurality of processors;
• executing the application by a host processor of the plurality of servers, wherein executing the application comprises (i) dispatching first kernel code to a second processor of the plurality of processors, wherein the first kernel code comprises computing functions of the application and (ii) dispatching second kernel code to the second I/O processor;
• executing, by the second processor, the first kernel code to perform the computing functions of the application;
• executing, by the first I/O processor, program code comprising network protocol functions and data storage protocol functions to perform data I/O processing functions and data storage functions for the application executing on the host processor, wherein the first I/O processor is configured to perform the data I/O processing functions and the data storage functions independent of the host processor;
• and
• executing, by the second I/O processor, the second kernel code to perform data I/O processing functions that are offloaded from the first I/O processor.

An article of manufacture comprising a non-transitory processor-readable storage medium having stored therein software instructions, which are executable by one or more hardware processor devices to:

• host an application on a server node, wherein the server node comprises:
• a plurality of processors comprising a first processor and a second processor;
• a first input/output (I/O) processor comprising an I/O hardware interface coupled to a front-end communications network and a backend data storage network, wherein the I/O hardware interface of the first I/O processor implements converged network adapter functions to provide a data entry point and data exit point for each of the front-end communications network and the backend data storage network;
• a second I/O processor;
• a first shared memory coupled to the plurality of processors;
• a second shared memory coupled to the first and second I/O processors;
• and
• a global memory coupled to the first shared memory and the second shared memory, wherein the global memory is configured to enable communication between the first and second I/O processors and the plurality of processors;
• execute the application by a host processor of the plurality of servers, wherein executing the application comprises (i) dispatching first kernel code to a second processor of the plurality of processors, wherein the first kernel code comprises computing functions of the application and (ii) dispatching second kernel code to the second I/O processor;
• execute, by the second processor, the first kernel code to perform the computing functions of the application;
• execute, by the first I/O processor, program code comprising network protocol functions and data storage protocol functions to perform data I/O processing functions and data storage functions for the application executing on the host processor, wherein the first I/O processor is configured to perform the data I/O processing functions and the data storage functions independent of the host processor;
• and
• execute, by the second I/O processor, the second kernel code to perform data I/O processing functions that are offloaded from the first I/O processor.