Preserving data of a volatile memory
Apparatuses, systems, methods, and computer program products for auto-commit memory are presented. A monitor module determines that a triggering event for an auto-commit memory has occurred. An identification module identifies a triggered commit action for an auto-commit memory. An auto-commit memory module performs a triggered commit action for an auto-commit memory in response to a triggering event occurring.
1. A method comprising:
storing data in a volatile memory communicatively coupled to a non-volatile memory device, the data associated with a logical identifier;
maintaining metadata associating the data with the logical identifier;
writing the data from the volatile memory to a non-volatile memory medium of the non-volatile memory device in response to a failure event such that the data on the non-volatile memory medium remains associated with the logical identifier; and
preventing a read before write hazard based on the maintained metadata, the metadata tracking associations between logical identifiers of a block device interface for the data and memory addresses for the data in the volatile memory.
2. The method of claim 1 , further comprising mapping the volatile memory into a volatile memory address space of a host computer in response to a request.
3. The method of claim 1 , further comprising mapping the volatile memory into a memory system of a remote computing device using remote direct memory access (RDMA) such that the data of the volatile memory is accessible on the remote computing device using memory semantics.
4. The method of claim 1 , further comprising disabling caching of memory operations for the volatile memory on a host computing device such that memory semantic operations for the volatile memory are performed directly on the volatile memory.
5. The method of claim 1 , further comprising clearing the metadata in response to writing the data to the non-volatile memory medium.
6. The method of claim 1 , wherein the logical identifier associated with the data is associated with a physical location on the non-volatile memory medium of a non-volatile storage device.
7. The method of claim 1 , wherein the non-volatile memory medium comprises a cache for a backing store and writing the data from the volatile memory to the non-volatile memory medium in response to the failure event preserves dirty data for migration from the cache to the backing store.
8. The method of claim 1 , wherein the data comprises a durable synchronization primitive, the volatile memory mapped into different address ranges of a memory system such that the synchronization primitive has a same state for each of the different address ranges.
9. The method of claim 1 , wherein the volatile memory is tightly coupled to a storage controller for the non-volatile memory medium, the storage controller configured to write the data from the volatile memory to the non-volatile memory medium independent of a host computing device.
10. An apparatus comprising:
a monitor module configured to determine that a triggering event for a volatile memory has occurred, the volatile memory mapped into a volatile memory system of a host computing device;
an identification module configured to identify a triggered commit action for the volatile memory;
an auto-commit memory module configured to perform the triggered commit action for the volatile memory during a power hold-up time provided by a secondary power source for the volatile memory in response to the monitor module determining that the triggering event has occurred; and
a storage layer configured to prevent a read before write hazard based on metadata maintained for data of the volatile memory, the metadata tracking associations between logical identifiers of a block device interface for the data of the volatile memory and memory addresses for the data of the volatile memory mapped into the volatile memory system of the host computing device, wherein the monitor module, the identification module, the auto-commit memory module, and the storage layer comprise one or more of logic hardware and executable code stored on a non-transitory computer readable storage medium.
11. The apparatus of claim 10 , further comprising an instant committed memory library configured to map the volatile memory into the volatile memory system of the host computing device in response to a request.
12. The apparatus of claim 10 , further comprising a storage layer configured to disable caching of memory operations for the volatile memory on a host computing device such that memory semantic operations for the volatile memory are performed directly on the volatile memory.
13. The apparatus of claim 10 , further comprising a storage layer configured to disarm the volatile memory by clearing the triggered commit action for the volatile memory in response to the auto-commit memory module performing the triggered commit action.
14. An apparatus comprising:
means for executing memory semantic operations on data in a volatile memory buffer mapped into a volatile memory address space of a host computer of a non-volatile storage device comprising the volatile memory buffer and disabling caching of the memory semantic operations on the host computer such that the memory semantic operations are executed directly on the volatile memory buffer;
means for arming the volatile memory buffer with metadata associating the data with a logical identifier of the non-volatile storage device; and
means for committing the data from the volatile memory buffer to a non-volatile memory medium of the non-volatile storage device in response to a trigger such that the logical identifier is stored together with the data associated with the logical identifier in the non-volatile memory medium, wherein the means for executing, the means for arming, and the means for committing comprise one or more of logic hardware and executable code stored on a non-transitory computer readable storage medium.
15. The apparatus of claim 14 , further comprising means for mapping the volatile memory buffer into the volatile memory address space of the host computer in response to a request.
16. The apparatus of claim 14 , further comprising means for preventing a read before write hazard based on the metadata, the metadata tracking associations between the logical identifier and memory addresses for the data of the volatile memory buffer.
17. The apparatus of claim 14 , further comprising means for disarming the volatile memory buffer by clearing the metadata in response to committing the data from the volatile memory buffer to the non-volatile storage device.