Failure fault tolerance in distributed memory systems

A memory processing unit for servicing read requests in a distributed memory system, the memory processing unit comprising:

• a host interface configured to communicate with a host;
• a memory fabric interface configured to communicate with a plurality of memory devices, the plurality of memory devices comprising a first memory device and a second memory device configured as a pair of memory devices;
• and
• a processor configured to perform operations comprising:
• receiving, from the host across the host interface, a first read request for a first memory location corresponding to the pair of memory devices;
• transmitting the first read request to the first memory device over the memory fabric interface;
• receiving a data-error response from the first memory device in response to the first read request;
• responsive to receiving the data-error response, transmitting the first read request to the second memory device over the memory fabric interface without first retrying the first memory device;
• returning, to the host, data received from the second memory device in response to the first read request;
• receiving, from the host across the host interface, a second read request for a second memory location corresponding to the pair of memory devices;
• transmitting the second read request to the first memory device over the memory fabric interface;
• detecting a communication failure associated with the first memory device in response to the second read request;
• responsive to detecting the communication failure, retrying the second read request to the first memory device over an alternative path before transmitting the second read request to the second memory device;
• and
• returning, to the host, data received in response to the second read request.

A non-transitory machine-readable medium storing instructions for servicing read requests in a distributed memory system, the instructions, when executed by a memory processing unit connected across a host interface to a host and a memory fabric interface to a plurality of memory devices, the plurality of memory devices comprising a first memory device and a second memory device configured as a pair of memory devices, cause the memory processing unit to perform operations comprising:

• receiving, from the host across the host interface, a first read request for a first memory location corresponding to the pair of memory devices;
• transmitting the first read request to the first memory device over the memory fabric interface;
• receiving a data-error response from the first memory device in response to the first read request;
• responsive to receiving the data-error response, transmitting the first read request to the second memory device over the memory fabric interface without first retrying the first memory device;
• returning, to the host, data received from the second memory device in response to the first read request;
• receiving, from the host across the host interface, a second read request for a second memory location corresponding to the pair of memory devices;
• transmitting the second read request to the first memory device over the memory fabric interface;
• detecting a communication failure associated with the first memory device in response to the second read request;
• responsive to detecting the communication failure, retrying the second read request to the first memory device over an alternative path before transmitting the second read request to the second memory device;
• and
• returning, to the host, data received in response to the second read request.