Systems and methods with integrated memory pooling and direct swap caching

A system comprising:

• a first compute node comprising a first local memory, coupled to a pooled memory system, wherein the pooled memory system comprises pooled memory;
• a first host operating system (OS), associated with the first compute node, having initial access to:
• (1) a first swappable range of memory addresses associated with the first local memory and a first non-swappable range of memory addresses associated with the first local memory, and (2) a second swappable range of memory addresses associated with the pooled memory;
• a second compute node comprising a second local memory, different from the first local memory, coupled to the pooled memory system;
• a second host operating system (OS), associated with the second compute node, having initial access to:
• (1) a third swappable range of memory addresses associated with the second local memory and a third non-swappable range of memory addresses associated with the second local memory, and (2) a fourth swappable range of memory addresses associated with the pooled memory;
• a first data-mover offload engine configured to, in response to a first slice of memory being taken offline with respect to the first host OS, perform a first cleanup operation, including:
• (1) restore a state of any memory content swapped-out from a memory location within the first swappable range of memory addresses to the pooled memory, and (2) move from the first local memory any memory content swapped-in from a memory location within the second swappable range of memory addresses back out to the pooled memory;
• and
• a second data-mover offload engine, different from the first data-mover engine, configured to, in response to a second slice of memory being taken offline with respect to the second host OS, perform a second cleanup operation, including:
• (1) restore a state of any memory content swapped-out from a memory location within the third swappable range of memory addresses to the pooled memory, and (2) move from the second local memory any memory content swapped-in from a memory location within the fourth swappable range of memory addresses back out to the pooled memory.

A method comprising:

• assigning to a compute node, comprising a local memory, pooled memory from within a pool of memory;
• granting to a host operating system (OS) access to:
• (1) a first swappable range of memory addresses associated with the local memory and a non-swappable range of memory addresses associated with the local memory, and (2) a second swappable range of memory addresses associated with the pooled memory;
• and
• using a dedicated data-mover offload engine, in response to a slice of memory being taken offline with respect to the host OS, automatically performing a cleanup operation, including:
• (1) restoring a state of any memory content swapped-out from a memory location within the first swappable range of memory addresses to the pooled memory, and (2) moving from the local memory any memory content swapped-in from a memory location within the second swappable range of memory addresses back out to the pooled memory.