Hierarchical storage device with host controlled subdivisions
This disclosure provides techniques hierarchical address virtualization within a memory controller and configurable block device allocation. By performing address translation only at select hierarchical levels, a memory controller can be designed to have predictable I/O latency, with brief or otherwise negligible logical-to-physical address translation time. In one embodiment, address transition may be implemented entirely with logical gates and look-up tables of a memory controller integrated circuit, without requiring processor cycles. The disclosed virtualization scheme also provides for flexibility in customizing the configuration of virtual storage devices, to present nearly any desired configuration to a host or client.
1. A solid state storage drive (SSD) comprising:
an interface that employs a non-volatile memory express (NVMe) command format to communicate with a storage manager of a host;
flash memory having physical planes, wherein the flash memory is logically organized so as to have block devices, and for a given one of the block devices, segments, each of the segments mapped to a respective, mutually-exclusive subset of erase units, including erase units that are on different ones of the physical planes, and each segment having a respective set of logical block addresses (LBAs); and
logic operable to cause the SSD to:
receive via the interface, from the storage manager of the host, write commands, wherein the write commands are directed to a specific one of the segments, wherein the write commands direct programming of data for a numerically-consecutive subset of LBAs from the set of LBAs which is respective to the specific one of the segments;
for each one of the numerically-consecutive LBAs:
derive from incoming host-provided address information a first address portion and a second address portion;
identify from the first address portion the given one of the block devices;
perform one or more operations to subdivide the second address portion to identify the specific one of the segments and a storage location within the addressed segment;
derive a common physical page address; and
program data corresponding to the numerically-consecutive subset of LBAs into the erase units that are on the different ones of the physical planes using the common physical page address, in a manner such that the programming of the data corresponding to the numerically-consecutive subset of LBAs is at least partially concurrent in time; and
receive via the interface, from the storage manager of the host, a maintenance command directed to a specific one of the segments, and in response to the maintenance command, physically erase each of the erase units in the respective, mutually-exclusive subset which is mapped to the specific one of the segments;
wherein said logic comprises at least one of hardware circuitry or instructions stored on at least one physical storage medium that, when executed, are to control operation of hardware circuitry.
2. The SSD of claim 1 , wherein the logic is further operable to cause the SSD to, during the erasure of the erase units which is responsive to the maintenance command, detect a failure condition of a specific one of the physical erase units which is both in the respective, mutually-exclusive subset which is mapped to the specific one of the segments and which is on a specific one of the physical planes and, in response to detection of the failure condition, map the specific segment to a new physical erase unit which is on the specific one of the physical planes, the new physical erase unit being distinct from the respective sets of the physical erase units, the new physical erase unit is mapped only to the specific segment, and not to any other one of the segments.
3. The SSD of claim 1 , wherein the logic is further operable to cause the SSD to:
store metadata which is respective to each of the segments;
unsolicitedly update the metadata as an automated response to data access operations in the SSD;
via the interface, receive a query command from the storage manager;
in response to the query command, identify information which is dependent on the unsolicitedly updated metadata and transmit the information to the storage manager; and
via the interface, receive and execute one or more commands, from the storage manager, which are dependent on the transmitted information.
4. The SSD of claim 3 , wherein the information transmitted to the storage manager indicates an extent to which the erase units in the respective, mutually-exclusive subset which is mapped to the specific one of the segments are full.
5. The SSD of claim 3 , wherein:
the metadata indicates an extent to which the erase units in the respective, mutually-exclusive subset which is mapped to the specific one of the segments are full;
the logic is operable to cause the SSD to detect, based on the metadata, when the extent satisfies a threshold condition; and
the information transmitted to the storage manager both identifies the specific one of the segments and indicates the threshold condition.
6. The SSD of claim 5 wherein the one or more commands comprise the maintenance command.
7. The SSD of claim 1 , wherein the logic is further operable to cause the SSD to:
store metadata which is respective to each of the segments;
unsolicitedly update the metadata as an automated response to data access operations in the SSD; and
transmit, to the storage manager, information which is dependent on the unsolicitedly updated metadata, wherein the transmitted information corresponds to a given one of the segments.
8. The SSD of claim 7 wherein the transmitted information explicitly identifies the given one of the segments, and indicates a need for maintenance for the given one of the segments.
9. The SSD of claim 7 , wherein the information transmitted to the storage manager indicates an extent to which the erase units in the respective, mutually-exclusive subset which is mapped to the given one of the segments are full.
10. The SSD of claim 7 , wherein:
the metadata indicates an extent to which the erase units in the respective, mutually-exclusive subset which is mapped to the given one of the segments are full;
the logic is operable to cause the SSD to detect, based on the metadata, when the extent satisfies a threshold condition; and
the information transmitted to the storage manager both identifies the given one of the segments and indicates the threshold condition.
11. The SSD of claim 7 wherein the metadata respective to the given one of the segments comprises a time measure.
12. The storage device of claim 1 , wherein:
the SSD further comprises at least one register to store a capacity, for at least one of the segments, representing a minimum amount of storage space that is to be physically erased; and
the logic is further operable to cause the SSD to:
transmit, to the storage manager, information which conveys the capacity;
via the interface, receive and execute at least one command from the storage manager; and
in response to the at least one command, physically erase the erase units in the respective, mutually-exclusive subset which is mapped to the at least one of the segments.
13. The SSD of claim 1 , wherein:
the SSD further comprises at least one register to store a capacity, for at least one of the segments; and
the capacity can be defined to be exactly equal to a non-power of two number of storage locations.
14. The SSD of claim 1 , wherein:
the interface is also to receive incoming data read requests which each specify corresponding read address information and which are targeted to the given one of the block devices and the specific one of the segments; and
the logic is operable to cause the SSD to, for each of the incoming data read requests:
derive, from the corresponding read address information, a first read address portion and a second read address portion;
identify from the first read address portion the given one of the block devices;
perform one or more operations to subdivide the second read address portion to identify the specific one of the segments and a read data storage location within the specific one of the segments; and
service the corresponding incoming data read request by retrieving read data from the read data storage location and transmitting the retrieved read data to the storage manager.
15. The SSD of claim 1 , wherein the logic is further operable to cause the SSD to:
store metadata which is respective to each of the segments;
unsolicitedly update the metadata as an automated response to data access operations in the SSD;
transmit, to the storage manager, information which is dependent on the unsolicitedly updated metadata, wherein the transmitted information corresponds to a given one of the segments; and
receive, via the interface, one or more commands, from the storage manager, and responsively reset metadata corresponding to the given one of the segments, as a function of executing the one or more commands.
16. The SSD of claim 1 , wherein:
the flash memory comprises flash memory dies;
the erase units in the respective, mutually-exclusive subset which is mapped to the specific one of the segments are in a number of the flash memory dies; and
for each corresponding one of the numerically-consecutive LBAs, the one or more operations include a division operation, performed on at least a portion of the corresponding one the numerically-consecutive LBAs, using an operand that is, at least in part, dependent on the number of flash memory dies, to derive therefrom the common page address.
17. The SSD of claim 1 , wherein at least one of the segments is mapped to erase units in at least four different ones of the physical planes.
18. An apparatus comprising:
a host having a storage manager; and
a solid state storage drive (SSD) comprising:
an interface that employs a non-volatile memory express (NVMe) command format to communicate with the storage manager;
flash memory having physical planes, wherein the flash memory is logically organized so as to have block devices, and for a given one of the block devices, segments, each of the segments mapped to a respective, mutually-exclusive subset of erase units, including erase units that are on different ones of the physical planes, and each segment having a respective set of logical block addresses (LBAs); and
logic operable to cause the SSD to:
receive via the interface, from the storage manager of the host, write commands, wherein the write commands are directed to a specific one of the segments, wherein the write commands direct programming of data for a numerically-consecutive subset of LBAs from the set of LBAs which is respective to the specific one of the segments;
for each one of the numerically-consecutive LBAs:
derive from incoming host-provided address information a first address portion and a second address portion;
identify from the first address portion the given one of the block devices;
perform one or more operations to subdivide the second address portion to identify the specific one of the segments and a storage location within the addressed segment;
derive a common physical page address; and
program data corresponding to the numerically-consecutive subset of LBAs into the erase units that are on the different ones of the physical planes using the common physical page address, in a manner such that the programming of the data corresponding to the numerically-consecutive subset of LBAs is at least partially concurrent in time; and
receive via the interface, from the storage manager of the host, a maintenance command directed to a specific one of the segments, and in response to the maintenance command, physically erase each of the erase units in the respective, mutually-exclusive subset which is mapped to the specific one of the segments;
wherein said logic comprises at least one of hardware circuitry or instructions stored on at least one physical storage medium that, when executed, are to control operation of hardware circuitry.
19. The apparatus of claim 18 , wherein the logic is further operable to cause the SSD to, during the erasure of the erase units which is responsive to the maintenance command, detect a failure condition of a specific one of the physical erase units which is both in the respective, mutually-exclusive subset which is mapped to the specific one of the segments and which is on a specific one of the physical planes and, in response to detection of the failure condition, map the specific segment to a new physical erase unit which is on the specific one of the physical planes, the new physical erase unit being distinct from the respective sets of the physical erase units, the new physical erase unit is mapped only to the specific segment, and not to any other one of the segments.
20. The apparatus of claim 18 , wherein the logic is further operable to cause the SSD to:
store metadata which is respective to each of the segments;
unsolicitedly update the metadata as an automated response to data access operations in the SSD;
via the interface, receive a query command from the storage manager;
in response to the query command, identify information which is dependent on the unsolicitedly updated metadata and transmit the information to the storage manager; and
via the interface, receive and execute one or more commands, from the storage manager, which are dependent on the transmitted information.
21. The apparatus of claim 18 , wherein the logic is further operable to cause the SSD to:
store metadata which is respective to each of the segments;
unsolicitedly update the metadata as an automated response to data access operations in the SSD; and
transmit, to the storage manager, information which is dependent on the unsolicitedly updated metadata, wherein the transmitted information corresponds to a given one of the segments.
22. The apparatus of claim 21 , wherein:
the metadata indicates an extent to which the erase units in the respective, mutually-exclusive subset which is mapped to the given one of the segments are full;
the logic is operable to cause the SSD to detect, based on the metadata, when the extent satisfies a threshold condition;
the information transmitted to the storage manager both identifies the given one of the segments and indicates the threshold condition; and
the storage manager is to transmit, to the SSD, at least one command in a manner responsive to the information transmitted to the storage manager by the SSD.