Controlled system management based on thermal load of storage device
By integrating the controller of the ZNS architecture in the data storage device, analyzing the thermal statistics of the die and selecting the appropriate die to form a partition, the problem of uneven thermal load of the enterprise-side data storage device is solved and the system performance is improved.
Patent Information
- Application Number
- CN202380014588.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-04-06
- Filing Date
- 2023-04-04
- Publication Date
- 2024-11-05
AI Technical Summary
Enterprise data storage devices have uneven temperatures due to high thermal loads, which affect performance. The existing thermal throttling measures will reduce performance, making it difficult to achieve effective load and thermal load balancing.
Design a storage system to collect and analyze thermal statistics of individual dies by integrating a controller with a partition namespace (ZNS) architecture in a data storage device, selecting appropriate dies to form partitions to achieve load and thermal load balance.
Through dynamic partition management and load balancing, the temperature of data storage devices is reduced, system performance is improved, and performance degradation caused by thermal load is avoided.
Smart Images

Figure CN118901052A_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of, and is hereby incorporated by reference in its entirety, U.S. Non-Provisional Application No. 17 / 714,863, filed on April 6, 2022, and entitled “Controlled System Management Based On Storage Device Thermal Load,” for all purposes. Background Art Technical Field
[0003] Embodiments of the present disclosure generally relate to storage systems having data storage devices, such as solid-state drives (SSDs) using a partitioned namespace architecture, and more particularly to programming data based on thermal statistics and load balancing of the storage systems and data storage devices.
[0004] Description of Related Art
[0005] Enterprise-side data storage devices typically have large capacities, which require multiple dies of large form factors. Since enterprise-side performance expectations are often line rate, enterprise-side data storage devices also require a lot of power, which can result in high thermal loads. The thermal load can be distributed between the dies in multiple dies. Because the workload may not be evenly distributed across the dies in multiple dies, some parts of the enterprise-side data storage device may be hotter than other parts of the enterprise-side data storage device based on workload and usage. For example, a die that experiences more use or a larger workload may have a higher temperature than a die that experiences less use or a smaller workload.
[0006] Additionally, due to die manufacturing variability, the power consumption of each die may be different, depending on which die is targeted in the workload. In addition, the physical proximity of each die to the airflow source may cause temperature differences to occur in multiple dies. Similarly, the location of each die may also affect adjacent dies. Therefore, the thermal load and potential throughput may be affected by the thermal distribution of enterprise-end data storage devices. In order to reduce the temperature of data storage devices that reach or exceed a threshold temperature, thermal throttling can be used. However, thermal throttling reduces the performance of the data storage device, which may cause the data storage device to fail to meet performance expectations.
[0007] Therefore, there is a need in the art to improve workload and thermal load balancing within storage systems and within data storage devices. Summary of the invention
[0008] The present disclosure generally relates to a storage system having a data storage device, such as a solid-state drive (SSD) using a partition namespace architecture, and more specifically to programming data based on thermal statistics and load balancing of the storage system and the data storage device. A storage system includes two or more data storage devices and a controller coupled to the two or more data storage devices. Each of the two or more data storage devices includes a partition namespace (ZNS) architecture. The controller is configured to collect thermal statistics of each of the two or more data storage devices, analyze the collected thermal statistics, and based on the analyzed collected thermal statistics, specify a partition by selecting one or more dies within at least one of the two or more data storage devices. The data storage device includes a memory device having a plurality of dies and a controller coupled to the memory device. The controller is configured to collect thermal statistics of each of the plurality of dies, analyze the collected statistics, and allocate one or more dies to form a partition.
[0009] In one embodiment, a storage system includes: two or more data storage devices, wherein each of the two or more data storage devices includes a zoned namespace (ZNS) architecture; and a controller coupled to the two or more data storage devices. The controller is configured to collect thermal statistics of each of the two or more data storage devices, analyze the collected thermal statistics, and specify a partition by selecting one or more dies within at least one of the two or more data storage devices based on the analyzed collected thermal statistics.
[0010] In another embodiment, a data storage device includes a memory device having a zoned name space (ZNS) architecture, wherein the memory device includes a plurality of dies, and a controller coupled to the memory device. The controller is configured to collect thermal statistics of each of the plurality of dies, analyze the collected statistics, and allocate one or more of the plurality of dies to form a partition.
[0011] In another embodiment, a storage system includes two or more memory devices and a controller coupled to the one or more memory devices. The controller is configured to allocate a partition formed by a portion of one or more memory devices of the two or more memory devices based on a thermal statistic of each of the two or more memory devices. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Thus, a detailed understanding of the manner in which the above-described features of the present disclosure are understood, a more particular description of the present disclosure, the above brief overview, may be obtained by reference to the embodiments, some of which are illustrated in the accompanying drawings. It should be noted, however, that the drawings illustrate only typical embodiments of the present disclosure and are therefore not to be considered limiting of its scope, as the present disclosure may admit to other equally effective embodiments.
[0013] Figure 1 is a schematic block diagram illustrating a storage system according to certain embodiments, in which a data storage device may be used as a storage device for a host device.
[0014] Figure 2 is an exemplary illustration of a storage system having multiple data storage devices according to certain embodiments.
[0015] Figure 3A is an exemplary illustration of a partitioned namespace utilized in a storage device according to certain embodiments.
[0016] Figure 3B According to certain embodiments Figure 3A An exemplary illustration of a state diagram of a partitioned namespace of a storage device.
[0017] Figure 4 is an exemplary illustration of forming partitions in a memory device according to certain embodiments.
[0018] Figure 5 is a flow chart illustrating a method of providing partitions to a host device according to certain embodiments.
[0019] Figure 6 is a schematic block diagram illustrating a storage system including a controller coupled to a plurality of data storage devices according to certain embodiments, wherein the controller includes a load balancer and a thermal statistics collector / analyzer module.
[0020] Figure 7 is a flow chart illustrating a method of rebalancing access queues of a data storage device according to certain embodiments.
[0021] Figure 8 is a flow chart illustrating a method of processing write commands in a storage system with load balancing based on thermal statistics and future workloads according to certain embodiments.
[0022] To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures. It is contemplated that elements disclosed in one embodiment may be beneficially utilized on other embodiments without specific recitation. DETAILED DESCRIPTION
[0023] Hereinafter, reference is made to embodiments of the present disclosure. However, it should be understood that the present disclosure is not limited to the specifically described embodiments. On the contrary, any combination of the following features and elements (whether or not related to different embodiments) is considered to realize and practice the present disclosure. In addition, although the embodiments of the present disclosure can achieve advantages that are superior to other possible solutions and / or superior to the prior art, whether a specific advantage is achieved by a given embodiment is not a limitation to the present disclosure. Therefore, the following aspects, features, embodiments and advantages are merely illustrative, and are not considered to be elements or limitations of the appended claims, unless explicitly stated in the claims. Similarly, reference to "the present disclosure" should not be interpreted as a summary of any inventive subject matter disclosed herein, and should not be considered to be elements or limitations of the appended claims, unless explicitly stated in the claims.
[0024] The present disclosure generally relates to a storage system having a data storage device, such as a solid-state drive (SSD) using a partition namespace architecture, and more specifically to programming data based on thermal statistics and load balancing of the storage system and the data storage device. A storage system includes two or more data storage devices and a controller coupled to the two or more data storage devices. Each of the two or more data storage devices includes a partition namespace (ZNS) architecture. The controller is configured to collect thermal statistics of each of the two or more data storage devices, analyze the collected thermal statistics, and based on the analyzed collected thermal statistics, specify a partition by selecting one or more dies within at least one of the two or more data storage devices. The data storage device includes a memory device having a plurality of dies and a controller coupled to the memory device. The controller is configured to collect thermal statistics of each of the plurality of dies, analyze the collected statistics, and allocate one or more dies to form a partition.
[0025] Figure 11 is a schematic block diagram illustrating a storage system 100 according to certain embodiments, the storage system having a data storage device 106 that can be used as a storage device for a host device 104. For example, the host device 104 can utilize a non-volatile memory (NVM) 110 included in the data storage device 106 to store and retrieve data. The host device 104 includes a host DRAM 138. In some examples, the storage system 100 may include multiple storage devices, such as the data storage device 106, that can operate as a storage array. For example, the storage system 100 may include multiple data storage devices 106 that are configured as a redundant array of inexpensive / independent disks (RAID) that are collectively used as a large-capacity storage device for the host device 104. The storage system 100 can be an enterprise-side storage system or a client-side storage system. Both the enterprise-side storage system and the client-side storage system include one or more data storage devices, wherein the enterprise-side storage system includes a data storage device greater than or equal to a threshold number, and the client-side storage system includes a data storage device less than a threshold number.
[0026] Host device 104 may store data to and / or retrieve data from one or more storage devices, such as data storage device 106. Figure 1 As shown, host device 104 can communicate with data storage device 106 via interface 114. Host device 104 can include any of a variety of devices, including a computer server, a network attached storage (NAS) unit, a desktop computer, a notebook (i.e., laptop) computer, a tablet computer, a set-top box, a telephone handset such as a so-called "smart" phone, a so-called "smart" tablet computer, a television, a camera, a display device, a digital media player, a video game console, a video streaming device, or other device capable of sending or receiving data from a data storage device.
[0027] Data storage device 106 includes controller 108, NVM 110, power supply 111, volatile memory 112, interface 114, and write buffer 116. In some examples, for clarity, data storage device 106 may include Figure 1. For example, the data storage device 106 may include a printed circuit board (PCB) to which the components of the data storage device 106 are mechanically attached, and the printed circuit board includes conductive traces that electrically interconnect the components of the data storage device 106, etc. In some examples, the physical size and connector configuration of the data storage device 106 may conform to one or more standard form factors. Some exemplary standard form factors include, but are not limited to, a 3.5" data storage device (e.g., an HDD or SSD), a 2.5" data storage device, a 1.8" data storage device, a peripheral component interconnect (PCI), a PCI extension (PCI-X), a PCI Express (PCIe) (e.g., PCIe x1, x4, x8, x16, PCIe Mini card, MiniPCI, etc.). In some examples, the data storage device 106 may be directly coupled (e.g., directly soldered or inserted into a connector) to a motherboard of the host device 104.
[0028] The interface 114 may include one or both of a data bus for exchanging data with the host device 104 and a control bus for exchanging commands with the host device 104. The interface 114 may operate according to any suitable protocol. For example, the interface 114 may operate according to one or more of the following protocols: Advanced Technology Attachment (ATA) (e.g., Serial ATA (SATA) and Parallel ATA (PATA)), Fibre Channel Protocol (FCP), Small Computer System Interface (SCSI), Serial Attached SCSI (SAS), PCI and PCIe, Non-Volatile Memory Express (NVMe), OpenCAPI, GenZ, Cache Coherent Interface Accelerator (CCIX), Open Channel SSD (OCSSD), etc. The interface 114 (e.g., a data bus, a control bus, or both) is electrically connected to the controller 108, thereby providing an electrical connection between the host device 104 and the controller 108, allowing data to be exchanged between the host device 104 and the controller 108. In some examples, the electrical connection of the interface 114 may also allow the data storage device 106 to receive power from the host device 104. For example, as Figure 1 As shown, power supply 111 may receive power from host device 104 via interface 114 .
[0029] NVM 110 may include multiple memory devices or memory cells. NVM 110 may be configured to store and / or retrieve data. For example, a memory cell of NVM 110 may receive data and a message from controller 108 indicating that the memory cell stores the data. Similarly, the memory cell may receive a message from controller 108 indicating that the memory cell retrieves data. In some examples, each of the memory cells may be referred to as a die. In some examples, NVM 110 may include multiple dies (e.g., multiple memory cells). In some examples, each memory cell may be configured to store a relatively large amount of data (e.g., 128MB, 256MB, 512MB, 1GB, 2GB, 4GB, 8GB, 16GB, 32GB, 64GB, 128GB, 256GB, 512GB, 1TB, etc.).
[0030] In some examples, each memory cell may include any type of non-volatile memory device, such as a flash memory device, a phase change memory (PCM) device, a resistive random access memory (ReRAM) device, a magnetoresistive random access memory (MRAM) device, a ferroelectric random access memory (F-RAM), a holographic memory device, and any other type of non-volatile memory device.
[0031] NVM 110 may include multiple flash memory devices or memory cells. NVM flash memory devices may include NAND or NOR-based flash memory devices, and may store data based on the charge contained in the floating gate of the transistor for each flash memory cell. In NVM flash memory devices, the flash memory device may be divided into multiple dies, wherein each of the multiple dies includes multiple physical or logical blocks, which may be further divided into multiple pages. Each of the multiple blocks within a particular memory device may include multiple NVM cells. The rows of NVM cells may be electrically connected using word lines to define pages in multiple pages. The corresponding cells in each of the multiple pages may be electrically connected to the corresponding bit lines. In addition, the NVM flash memory device may be a 2D or 3D device, and may be a single-level cell (SLC), a multi-level cell (MLC), a three-level cell (TLC), or a four-level cell (QLC). The controller 108 may write data to and read data from the NVM flash memory device at the page level, and erase data from the NVM flash memory device at the block level.
[0032] The power supply 111 can provide power to one or more components of the data storage device 106. When operating in standard mode, the power supply 111 can use power provided by an external device such as the host device 104 to power one or more components. For example, the power supply 111 can use power received from the host device 104 via the interface 114 to power one or more components. In some examples, the power supply 111 may include one or more power storage components that are configured to power one or more components when operating in a shutdown mode, such as when power is stopped from being received from an external device. In this way, the power supply 111 can be used as an onboard backup power supply. Some examples of one or more power storage components include, but are not limited to, capacitors, supercapacitors, batteries, and the like. In some examples, the amount of power that can be stored by one or more power storage components can be a function of the cost and / or size (e.g., area / volume) of such one or more power storage components. In other words, as the amount of power stored by one or more power storage components increases, the cost and / or size of the one or more power storage components also increases.
[0033] Volatile memory 112 may be used by controller 108 to store information. Volatile memory 112 may include one or more volatile memory devices. In some examples, controller 108 may use volatile memory 112 as a cache. For example, controller 108 may store cached information in volatile memory 112 until the cached information is written to NVM 110. Figure 1 As shown, the volatile memory 112 may consume power received from the power supply 111. Examples of the volatile memory 112 include, but are not limited to, random access memory (RAM), dynamic random access memory (DRAM), static RAM (SRAM), and synchronous dynamic RAM (SDRAM (e.g., DDR1, DDR2, DDR3, DDR3L, LPDDR3, DDR4, LPDDR4, etc.)).
[0034] The controller 108 can manage one or more operations of the data storage device 106. For example, the controller 108 can manage reading data from the NVM 110 and / or writing data to the NVM. In some embodiments, when the data storage device 106 receives a write command from the host device 104, the controller 108 can initiate a data storage command to store data to the NVM 110 and monitor the progress of the data storage command. The controller 108 can determine at least one operating characteristic of the storage system 100 and store the at least one operating characteristic to the NVM 110. In some embodiments, when the data storage device 106 receives a write command from the host device 104, the controller 108 temporarily stores the data associated with the write command in an internal memory or write buffer 116 before sending the data to the NVM 110.
[0035] The controller includes a thermal statistics collector module 150. The thermal statistics collector module 150 is coupled to the NVM 110. The thermal statistics collector module 150 can be configured to retrieve thermal statistics of the NVM 110, wherein the thermal statistics can be the temperature of each of the multiple partitions of the NVM 110. In one embodiment, each partition can be a different memory device of the NVM 110, wherein the different memory devices have different memory architectures (e.g., SLC memory, MLC memory, TLC memory, QLC memory, PLC memory, etc.). In another embodiment, each partition can include two or more adjacent dies of the multiple dies of the NVM 110. In yet another embodiment, each partition includes one of the multiple dies of the NVM 110. The thermal statistics collector module 150 can be configured to generate a temperature distribution or heat map of the NVM 110.
[0036] Figure 2 is an exemplary illustration of a storage system 200 having a plurality of data storage devices according to certain embodiments, wherein each of the plurality of data storage devices may be Figure 1Data storage device 106. Multiple data storage devices are represented as points in a coordinate system, where each point (e.g., (1,1)) represents a data storage device in the storage system 200. The storage system 200 may include any number of data storage devices, such that it may include "x" times "y" number of data storage devices. The data storage devices in the storage system 200 may be arranged in rows and columns. It should be understood that the depicted configuration is not intended to be limiting, but is provided only as an exemplary embodiment. For example, the storage system may have a configuration mode other than rows and columns. It should also be understood that although a planar arrangement of data storage devices is shown, the storage system 200 may have a stacked configuration. In other words, data storage devices may exist above and / or below the depicted exemplary illustration of the storage system 200.
[0037] Each data storage device is separated from adjacent data storage devices to allow airflow from airflow sources 202, physical components (e.g., wiring), etc. For example, a data storage device located at (1,1) is at a distance "a" from an adjacent data storage device located at (1,2), at a distance "b" from an adjacent data storage device located at (2,1), and at a distance "c" from a diagonally adjacent data storage device located at (2,2). It should be understood that "a", "b", and "c" may be the same or different. Similarly, in embodiments where data storage devices are stacked, the distance between adjacent stacked data storage devices is considered. In the storage system 200, there is a temperature distribution between each of the data storage devices in the data storage devices. The temperature distribution may depend on several factors, such as the distance between adjacent data storage devices (e.g., "a", "b", and "c"), the proximity of the data storage device to the center of the storage system 200, the workload of the data storage device, the distance of the data storage device from the airflow source 202, etc. For example, a data storage device at the center of the storage system 200 may dissipate heat more slowly than a data storage device toward the edge of the storage system 200. Likewise, data storage devices that are closer to the airflow source 202 may be cooler than data storage devices that are farther from the airflow source 202. In another example, data storage devices with higher workloads may be hotter than data storage devices with lower workloads. Using a fixed benchmark run on each of the data storage device operations in the data storage device operations, the thermal relationships between adjacent data storage devices may be predetermined or derived. In one example, analyzing the collected thermal statistics occurs during an offline storage system management period, such as when the storage system 200 is idle.
[0038] Figure 3Ais an exemplary illustration of a zoned namespace (ZNS) 302 utilized in a data storage device 300 according to certain embodiments. The data storage device 300 may present a view of the ZNS 302 to a host device, such as Figure 1 The storage device 300 may be a host device 104. Figure 1 The data storage device 106 of the storage system 100 of the present invention. The data storage device 300 may have one or more ZNS 302, and the size of each ZNS 302 may be different. In addition to the one or more partition namespaces 302, the data storage device 300 may also include one or more regular namespaces. In addition, the ZNS 302 may be a zone block command (ZBC) for SAS and / or a zone device ATA command set (ZAC) for SATA. Due to the relationship between possible logical and physical activities, host-side partition activities may be more directly related to media activities in the partition driver.
[0039] In data storage device 300, ZNS 302 is a quantity of NVM that can be formatted into logical blocks such that the capacity is divided into a plurality of partitions 306a through 306n (collectively, partitions 306). NVM can be Figure 1 10. Each of the partitions 306 includes a plurality of physical blocks or erase blocks (not shown) of the memory unit or NVM 304, and each of the erase blocks is associated with a plurality of logical blocks (not shown). The size of each of the partitions 306 may be consistent with the capacity of one or more erase blocks of the NVM or NAND device. When the controller 308 receives a command, such as from a host device (not shown) or a submission queue of the host device, the controller 308 may read data from and write data to a plurality of logical blocks associated with a plurality of erase blocks (EBs) of the ZNS 302. Each of the logical blocks is associated with a unique LBA or sector.
[0040] In one embodiment, NVM 304 is a NAND device. The NAND device includes one or more dies. Each of the one or more dies includes one or more planes. Each of the one or more planes includes one or more erase blocks. Each erase block in the one or more erase blocks includes one or more word lines (e.g., 256 word lines). Each word line in the one or more word lines can be addressed in one or more pages. For example, an MLC NAND die can use an upper page and a lower page to reach two bits in each cell of the entire word line (e.g., 16KiB per page). In addition, each page can be accessed with a granularity equal to or less than a complete page. The controller can frequently access the NAND with a user data granularity logical block address (LBA) size of 512 bytes. Therefore, as mentioned in the description below, the NAND position is equal to the granularity of 512 bytes. Therefore, the LBA size is 512 bytes and the page size of two pages of the MLC NAND is 16KiB, which results in 32 LBAs per word line. However, the NAND position size is not intended to be limited and is only used as an example.
[0041] When data is written to an erase block, one or more logical blocks are correspondingly updated within a partition 306 to track the location of the data within the NVM 304. Data can be written one partition 306 at a time until the partition 306 becomes full, or written to multiple partitions 306 so that multiple partitions 306 can become partially full. Similarly, when data is written to a particular partition 306, the data can be written to multiple erase blocks one block at a time in the order of NAND positions, page by page, or word line by word line until moving to an adjacent block (i.e., writing to the first erase block until the first erase block becomes full before moving to the second erase block), or can be written to multiple erase blocks multiple blocks at a time in the order of NAND positions, page by page, or word line by word line to partially fill each block in a parallel manner (i.e., writing to the first NAND position or page of each erase block before writing to the second NAND position or page of each erase block). This sequential programming of each NAND position is a typical non-limiting requirement of many NAND EBs.
[0042] When controller 308 selects to store the erase block of the data of each sub-region, controller 308 will be able to select erase block at the sub-region open time, or this controller can select erase block when satisfying the needs of the first word line of filling this specific erase block.When utilizing the above-mentioned method of filling an erase block completely before starting the next erase block, this may be more different.Controller 308 can use time difference to select more optimized erase block instantly.Which erase block is distributed and assigned to the decision of each sub-region and continuous LBA thereof can occur in controller 308 for zero or more parallel sub-regions all the time.
[0043] Each partition in the partitions 306 is associated with a partition starting logical block address (ZSLBA) or a partition starting sector. The ZSLBA is the first available LBA in the partitions 306. For example, the first partition 306a is associated with the ZSLBA. a SLBA is associated with the second partition 306b and Z b SLBA is associated with the third partition 306c and Z c SLBA is associated with the fourth partition 306d and Z d SLBA is associated, and the nth partition 306n (ie, the last partition) is associated with Z n Each partition 306 is identified by its ZSLBA and is configured to receive sequential writes (ie, data is written to NVM 110 in the order in which the write commands are received).
[0044] When data is written to a partition 306, a write pointer 310 is advanced or updated to point to or indicate the next available block in the partition 306 to which the data is written so as to track the next write start point (i.e., the completion point of the previous write is equal to the start point of the subsequent write). Thus, the write pointer 310 indicates where the subsequent write to the partition 306 will begin. Subsequent write commands are "partition append" commands, in which the data associated with the subsequent write command is appended to the partition 306 at the location indicated by the write pointer 310 as the next start point. An ordered list of LBAs within the partition 306 may be stored for write ordering. Each partition 306 may have its own write pointer 310. Thus, when a write command is received, the partition is identified by its ZSLBA, and the write pointer 310 determines the location at which the writing of the data begins within the identified partition.
[0045] Figure 3B According to certain embodiments Figure 3A 300 . In the state diagram 350 , each partition can be in a different state, such as an empty state, an active state, a full state, or an offline state. When a partition is empty, the partition contains no data (i.e., no erase blocks in the partition currently store data) and the write pointer is located at ZSLBA (i.e., WP=0). Once a write is scheduled to the partition or a partition open command is issued by the host, the empty partition will switch to an open and active partition. Partition management (ZM) commands can be used to move partitions between partition open and partition closed states (both are active states). If the partition is active, the partition includes open blocks that can be written to and a description of the recommended time in the active state can be provided to the host. The controller 308 includes the ZM. Partition metadata can be stored in the ZM and / or the controller 308.
[0046] The term "write" includes programming user data on 0 or more NAND locations in an erase block and / or partially filled NAND locations in an erase block when the user data has not yet filled all available NAND locations. The term "write" can also include moving a partition to a full state due to internal drive processing needs (open block data retention issues due to faster accumulation of bit errors on open erase blocks), the data storage device 300 closing or filling a partition due to resource limitations (like too many open partitions to track or defective conditions that have been discovered, etc.), or the host device closing a partition due to issues such as no more data to send to the drive, computer shutdown, error handling on the host, limited host resources for tracking, etc.
[0047] Active partitions can be open or closed. An open partition is an empty partition or partially filled partition that is ready for writing and has currently allocated resources. Data received from a host device using a write command or a partition append command can be programmed into an open erase block that is not currently filled with previous data. A closed partition is an empty partition or partially filled partition that is not currently receiving writes from the host continuously. Moving a partition from an open state to a closed state allows the controller 308 to reallocate resources to other tasks. These tasks may include, but are not limited to, other open partitions, other conventional non-partitioned areas, or other controller needs.
[0048] In both open and closed partitions, the write pointer points to a location in the partition between the ZSLBA and the end of the last LBA of the partition (i.e., WP>0). An active partition can switch between open and closed states as specified by the ZM, or when a write is scheduled to the partition. In addition, the ZM can reset the active partition to clear or erase the data stored in the partition, causing the partition to switch back to an empty partition. Once the active partition is full, the partition switches to a full state. A full partition is a partition that is completely filled with data and has no more available sectors or LBAs for writing data (i.e., WP=Partition Capacity (ZCAP)). In a full partition, the write pointer points to the end of the writable capacity of the partition. Read commands for data stored in a full partition can still be executed.
[0049] The partitions may have any total capacity, such as 256MiB, 512MiB, or 2MiB. However, a small portion of each partition may not be accessible to write data, but may still be read, such as a portion of each partition that stores parity data and one or more excluded erase blocks. For example, if the total capacity of partition 206 is 512MiB, the ZCAP may be 470MiB, which is the capacity available for writing data, while 42MiB is not available for writing data. In another example, if the total capacity of partition 306 is 2GiB, the ZCAP may be 1077MiB. The writable capacity (ZCAP) of the partition is equal to or less than the total partition storage capacity. The data storage device 300 may determine the ZCAP of each partition when the partition is reset. For example, the controller 308 or ZM may determine the ZCAP of each partition. When a partition is reset, the data storage device 300 may determine the ZCAP of the partition. It should be understood that the total capacity and ZCAP values listed previously are not intended to be limiting, but rather to provide examples of possible implementations.
[0050] The ZM can reset a full partition, thereby scheduling the erasure of the data stored in the partition so that the partition switches back to an empty partition. When resetting a full partition, although the partition may be marked as an empty partition available for writing, the data of the partition may not be cleared immediately. However, the reset partition must be erased before switching to an open and active partition. The partition can be erased at any time between the ZM reset and the ZM open. When resetting a partition, the data storage device 300 can determine the new ZCAP to reset the partition and update the writable ZCAP attribute in the partition metadata. An offline partition is a partition to which data cannot be written. An offline partition can be in a full state, an empty state, or a partially full state without being in an active state.
[0051] Since resetting a partition clears all data stored in the partition or schedules the erasure of all data stored in the partition, the need for garbage collection of each erase block is eliminated, thereby improving the overall garbage collection process of the data storage device 300. The data storage device 300 can mark one or more erase blocks for erasure. When a new partition is to be formed and the data storage device 300 expects that ZM is open, one or more erase blocks marked for erasure can be erased. The data storage device 300 can further determine and create the physical support of the partition when erasing the erase block. Therefore, once the new partition is opened and the erase block is selected to form the partition, the erase block will be erased. In addition, each time the partition is reset, the new order of the LBA and write pointer 310 of the partition 306 can be selected, so that the partition 306 can tolerate receiving commands out of order. The write pointer 310 can be optionally closed so that the command can be written to any starting LBA indicated by the command.
[0052] Re-reference Figure 3A , when the host device 104 sends a write command to write data to the partition 306, the controller 308 pulls in the write command and identifies the write command as writing to the newly opened partition 306. The controller 308 selects a set of EBs to store the data associated with the write command of the newly opened partition 306, and the newly opened partition 306 is switched to the active partition 306. The write command can be a command to write new data, or a command to move valid data to another partition for garbage collection purposes. The controller 308 is configured to DMA read new commands from the submission queue filled by the host device.
[0053] In the empty partition 306 that has just been switched to the active partition 306, because the write pointer 310 indicates the logical block associated with ZSLBA as the first available logical block, data is assigned to the partition 306 and a set of associated sequential LBAs of the partition 306 starting at ZSLBA. The data can be written to one or more erase blocks or NAND locations that have been allocated for the physical location of the partition 306. After the data associated with the write command has been written to the partition 306, the write pointer 310 is updated to point to the next LBA available for host writing (i.e., the completion point of the first write). The write data from the host write command is sequentially programmed into the next available NAND location in the erase block that is selected for physical support of the partition.
[0054] For example, the controller 308 may receive a first write command to the third partition 306c, or a first partition append command. The host device 104 sequentially identifies which logical block of the partition 306 is used to write the data associated with the first command. The data associated with the first command is then written to the first or next one or more available LBAs in the third partition 306c as indicated by the write pointer 310, and the write pointer 310 is advanced or updated to point to the next available LBA (i.e., WP>0) available for host writing. If the controller 308 receives a second write command to the third partition 306c, or a second partition append command, the data associated with the second write command is written to the next one or more available LBAs in the third partition 306c identified by the write pointer 310. Once the data associated with the second command is written to the third partition 306c, the write pointer 310 is again advanced or updated to point to the next available LBA available for host writing. Resetting the third partition 306c moves the write pointer 310 back to Z. c SLBA (ie, WP=0), and the third partition 306c switches to an empty partition.
[0055] In the description herein, for brevity and clarity, the term “erase block” may be referred to as a “block”.
[0056] Figure 4According to some embodiments, partitions are formed in the memory device 400 (such as Figure 3A 406a). The memory device 400 includes a plurality of dies 402a to 402n (collectively referred to as dies 402), wherein each of the plurality of dies 402a to 402n includes a first plane 404a and a second plane 404b (collectively referred to as plane 404). It should be understood that each die includes more than two planes (e.g., 4 planes, 8 planes, etc.). It should be understood that the embodiments herein may be applicable to any die architecture having one or more planes. Each plane 404 includes a plurality of blocks 406a to 406n (collectively referred to as blocks 406). Although 32 dies 402 are shown in the memory device 400, any number of dies may be included.
[0057] Each die may be separated from an adjacent die to allow airflow from an airflow source 408, a physical component (e.g., wiring), etc. For example, die 410a is at a distance "d" from adjacent die 410b, at a distance "e" from adjacent die 410c, and at a distance "f" from a diagonally adjacent die 410d. It should be understood that "d", "e", and "f" may be the same or different. Similarly, in embodiments where the die is stacked, the distance between adjacent stacked dies is considered. In the memory device 400, there is a temperature distribution between each of the plurality of dies 402a to 402n. The temperature distribution may depend on several factors, such as the distance between adjacent dies (e.g., "d", "e", and "f"), the proximity of the die to the center of the memory device 400, the workload of the die, the distance of the die from the airflow source 408, etc. For example, a die in the center of the memory device 400 may dissipate heat more slowly than a die toward the edge of the memory device 400. Likewise, a die that is closer to the airflow source 408 may be cooler than a die that is further from the airflow source 408. In another example, a die with a higher workload may be hotter than a die with a lower workload. Using analysis of the temperature of each die and a fixed benchmark used, thermal relationships between adjacent die may be predetermined or derived. In one example, analysis of the collected thermal statistics occurs during an offline storage system management period, such as when the memory device 400 is idle.
[0058] Such as Figure 3APartitions such as the first partition 306a of the memory device 400 include a block 406 from each plane 404 of each die 402. In some examples, a partition may include one or more blocks 406 from each plane 404 of each die 402. Additionally, in some embodiments, one or more of the dies 402 of the memory device 400 may be provided for storing XOR or parity data. The partitions may be formed based on the physical temperature distribution of the die 402 such that the blocks selected from the die are based on reducing the overall temperature increase of the memory device 400.
[0059] In addition, data is written sequentially from block to block in the first partition such that data is written to B0 406a before data is written to B1 406b. Data is also written sequentially from partition to partition such that data is written from the first partition before data is written to the second partition. The partitions may have any writable capacity (ZCAP), such as 256 MiB or 512 MiB, as discussed above. Each of the multiple partitions may have the same partition capacity. When, for example, Figure 1 When a data storage device such as the data storage device 106 receives a partition reset request (or in some cases, generates a partition reset request as part of a data management operation such as garbage collection), data of the partition capacity size is erased. In other words, unless the entire partition is erased or moved to a partition empty state (i.e., partition empty), individual blocks cannot be erased, such as Figure 3B However, if the data storage device 106 includes a non-volatile memory with partial ZNS functionality, then the partition capacity-sized data is erased from the data storage device 106 in the ZNS-enabled portion of the non-volatile memory. The block-sized data can be erased from the non-volatile memory unit without ZNS functionality.
[0060] In addition, data is stored in NVM (such as Figure 1 The location of the ZNS-capable portion of the NVM 110 is recorded in a first logical-to-physical (L2P) table as an LBA in a volatile memory unit (such as volatile memory 112). Data is stored in the NVM (such as Figure 1 The location in the non-ZNS-capable portion of NVM 110 of the present invention is recorded in the second L2P table as an LBA in a volatile memory unit (such as volatile memory 112). Volatile memory 112 may be a DRAM unit. In addition, NVM 110 may include a first L2P table that matches the first L2P table of volatile memory 112 and a second L2P table that matches the second L2P table of volatile memory 112. The L2P table in NVM 110 is updated to match the L2P table of volatile memory 112.
[0061] The L2P table may include a pointer to each physical location of data within NVM 110. The physical location of the data is mapped in the logical array so that the pointer address array includes locations mapped from the die to the NAND locations. In a block, the total number of pointers is calculated as follows: 256 WLs * 3 pages / WLs * 4 slots / page * 1 pointer / slot = 3,072 pointers. In a first partition with a capacity of 62 blocks, there may be 190,464 pointers (i.e., 3,072 pointers / block * 62 blocks = 190,464 pointers). Each pointer includes a certain amount of data using the available storage of volatile memory 112 and / or NVM 110. Although TLC memory is illustrated, the embodiments herein are not limited to TLC memory, and the embodiments described herein may be applicable to other memory densities, such as SLC memory, MLC memory, QLC memory, PLC memory, etc.
[0062] Figure 5 is a diagram showing a method of transmitting data to a host device (such as Figure 1 The host device 104 provides partitions (such as Figure 3A Flowchart of method 500 of first partition 306a of the system. Method 500 may be performed by a controller such as Figure 1 For exemplary purposes, reference may be made here to Figure 4 The memory device 400 and Figure 1 Aspects of the storage system 100 of the present invention. At box 502, the controller 108 receives a partition open command from the host device 104. The partition open command can be received explicitly (e.g., specified with the command) or implicitly (e.g., the command does not include a partition open command, but the command is directed to a partition that has not yet been opened). At box 504, the controller 108 analyzes the die thermal statistics. For example, the controller 108 can use the thermal statistics collector module 150 to collect thermal statistics for each die and analyze each die so that a thermal distribution of the die can be generated. At box 506, the controller 108 determines which group of dies to assign the partition to. At box 508, the controller 108 provides the newly opened partition to the host device 104.
[0063] In another embodiment, the physical location of the regular namespace can be selected based on the die temperature distribution. Regular namespace is a non-ZNS namespace in a ZNS device. Regular namespace can be a very small area where random writes are allowed. Regular namespaces can be located in die groups (much like partitions) and can be moved between die groups based on the thermal statistics of the die.
[0064] Figure 66 is a schematic block diagram showing a storage system 600 including a controller 606 coupled to a plurality of data storage devices 614a to 614n according to some embodiments, wherein the controller 606 includes a load balancer 608 and a thermal statistics collector / analyzer module 610 (which may be separate components). The controller 606 also includes a flash interface module (FIM) 612. In one example, the thermal statistics collector / analyzer module 610 may be disposed in the FIM 612. The storage system 600 may be Figure 2 The storage system 200 of FIG. 10 is a storage system 200 in which a controller 606 manages input / output traffic from one or more host devices to a plurality of data storage devices 614a to 614n. Each of the data storage devices 614a to 614n may be Figure 1 The data storage device 106. It should be understood that the one or more host devices may not be directly coupled, but linked via a network structure.
[0065] The controller 606 is coupled to the host device 602, where the host device 602 includes a command queue 604. The host device 602 generates commands and stores the generated commands in the command queue 604. In some examples, the host device 602 can send an indication to the controller 606 to obtain the generated commands stored in the command queue 604. In other examples, the host device 602 can send the generated commands from the command queue 604 to the controller 606.
[0066] The load balancer 608 may be configured to balance the future workload and thermal load of each of the data storage devices 614a to 614n. In addition, the thermal statistics collector / analyzer module 610 may be configured to collect and analyze thermal statistics from each of the multiple data storage devices 614a to 614n. For example, the analysis may also include determining the thermal impact of assigning commands from the command queue 604 to the data storage device, and how assigning commands from the command queue 604 to the data storage device may affect the temperature distribution of adjacent data storage devices. Based on the results of the thermal statistics collector / analyzer module 610, the load balancer 608 sends the received command from the command queue 604 to the FIM 612, which passes the command to the target data storage device.
[0067] In one example, each of the plurality of data storage devices 614a to 614n may further include a thermal statistics collector / analyzer module, such that the thermal statistics collector / analyzer module of each data storage device sends feedback to the host device 602. The host device 602 may modify the command queue 604 based on the feedback by adding or removing jobs or commands based on the thermal load of each data storage device.
[0068] Figure 7 To illustrate rebalancing of data storage devices (such as Figure 6 Flow chart of a method 700 for accessing a queue of one of a plurality of data storage devices 614a to 614n. The method 700 may be performed by Figure 6 For exemplary purposes, reference may be made here to Figure 6 At block 702, the controller 606 receives a future workload from the host device 602. At block 704, the controller 606 receives current thermal statistics from the thermal statistics collector / analyzer module 610. At block 706, the controller 606 rebalances access queues of the plurality of data storage devices 614a to 614n based on the workload and the current thermal statistics.
[0069] Rebalancing the access queues of multiple data storage devices 614a to 614n can be accomplished using different methods based on optimization techniques. For example, the optimization can follow a greedy schedule in which the thermal requirements of each data storage device are met one by one. When the current schedule does not meet the thermal requirements of the data storage device, the data storage device is pushed to the next time slot in the access queue that meets the thermal requirements of the data storage device. The controller 606 can also check that no other previously met thermal requirements of the data storage device are unmet by pushing the data storage device in the access queue. The controller 606 can also consider waiting time priority so that a relatively "hot" data storage device may not be the last data storage device to be accessed. For example, a token-based system can be used to implement waiting time priority so that each data storage device waiting in the access queue receives a token that increases its priority in the next time slot.
[0070] Figure 8 FIG. 1 is a diagram showing a method for performing a load balancing operation on a storage system (such as a storage system) based on thermal statistics and future workload according to certain embodiments. Figure 6 Flow chart of a method 800 for processing a write command in a storage system 600 of FIG. 1 . For exemplary purposes, reference may be made here to FIG. Figure 6 The storage system 600 and Figure 1 For example, the method 800 may be implemented by the controller 606.
[0071] At box 802, the controller 606 receives a write command from the host device 602 to write data to a partition of a data storage device in the plurality of data storage devices 614a to 614n. At box 804, the controller 606 collects thermal statistics of each of the plurality of data storage devices 614a to 614n. In some examples, the collection / collection of thermal statistics occurs continuously. At box 806, the controller 606 determines the current workload of each of the plurality of data storage devices 614a to 614n. At box 808, the controller 606 selects a data storage device in the plurality of data storage devices 614a to 614n based on the thermal statistics and the workload of each data storage device. The selected data storage device may be based on a thermal demand optimization scheme, such as the rebalancing scheme described above.
[0072] At box 810, the controller 606 determines whether there is an open partition in the selected data storage device. If there is an open partition in the selected data storage device at box 810, then at box 818, the controller 606 sends a write command to the controller of the data storage device, wherein the controller of the data storage device programs the data of the write command to the open partition. However, if there is no open partition in the data storage device selected at box 810, then at box 812, the controller 606 sends a partition open command to the selected data storage device. At box 814, the controller of the selected data storage device collects thermal statistics of multiple tube cores. At box 816, the controller of the selected data storage device opens a new partition based on the thermal statistics of the multiple tube cores. At box 818, the controller 606 sends a write command to the controller of the data storage device, wherein the controller of the data storage device programs the data of the write command to the open partition.
[0073] System performance of a storage system may be improved by balancing workloads and opening new partitions based on thermal statistics of each data storage device in a storage system including multiple data storage devices and multiple dies within each memory device of each data storage device.
[0074] In one embodiment, a storage system includes: two or more data storage devices, wherein each of the two or more data storage devices includes a zoned namespace (ZNS) architecture; and a controller coupled to the two or more data storage devices. The controller is configured to collect thermal statistics of each of the two or more data storage devices, analyze the collected thermal statistics, and specify a partition by selecting one or more dies within at least one of the two or more data storage devices based on the analyzed collected thermal statistics.
[0075] The controller is further configured to receive a partition open command and program host data to the specified partition. Receiving the partition open command occurs before analyzing the collected thermal statistics. Collecting the thermal statistics occurs continuously. Analyzing the collected thermal statistics occurs during an offline storage system management period. The selected one or more dies have a first thermal statistic and the remaining one or more dies have a second thermal statistic. The first thermal statistic is less than the second thermal statistic. The first thermal statistic and the second thermal statistic are temperature distributions. The controller is further configured to select the one or more dies based on a predicted workload. The two or more data storage devices are arranged in rows and columns to allow airflow therebetween. A first data storage of the two or more data storage devices has a first thermal statistic, and a second data storage device of the two or more data storage devices has a second thermal statistic. The first thermal statistic is different from the second thermal statistic. The first data storage device is located at a first distance from an airflow source, and the second data storage device is located at a second distance from an airflow source. The first distance and the second distance are different. Analyzing the collected thermal statistics includes analyzing the thermal statistics of the first data storage device and at least one of the thermal statistics of the second data storage device and the thermal statistics of the third data storage device. The second data storage device is arranged to be directly adjacent to the first data storage device, and the third data storage device is arranged to be diagonally adjacent to the first data storage device. The second data storage device is a plurality of data storage devices, and the third data storage device is a plurality of data storage devices. Each of the two or more data storage devices includes a plurality of tube cores. The controller is configured to select one or more tube cores of the plurality of tube cores to form the partition based on the collected thermal statistics of the plurality of tube cores.
[0076] In another embodiment, a data storage device includes a memory device having a zoned name space (ZNS) architecture, wherein the memory device includes a plurality of dies, and a controller coupled to the memory device. The controller is configured to collect thermal statistics of each of the plurality of dies, analyze the collected statistics, and allocate one or more of the plurality of dies to form a partition.
[0077] The thermal statistics of each of the plurality of dies are based at least on usage of the die, proximity of the die to an airflow source of the data storage device, and a temperature of a die adjacent to the die. The controller is further configured to provide the collected thermal statistics to a host device; based on the provided collected thermal statistics, receive a workload; and based on the received workload and the analyzed collected thermal statistics, rebalance an access queue of the data storage device. The controller is further configured to receive a workload type indication from a host device; receive the thermal statistics from a thermal statistics acquisition module, wherein the thermal statistics acquisition module is disposed in the controller; and based on the received workload type indication and the analyzed collected thermal statistics, rebalance an access queue of the data storage device.
[0078] In another embodiment, a storage system includes two or more memory devices and a controller coupled to the one or more memory devices. The controller is configured to allocate a partition formed by a portion of one or more memory devices of the two or more memory devices based on a thermal statistic of each of the two or more memory devices.
[0079] This storage system is an enterprise-side storage system. This storage system is a client-side storage system.
[0080] While the foregoing is directed to embodiments of the present disclosure, other and further embodiments of the present disclosure may be envisaged without departing from the basic scope of the present disclosure, and the scope of the present disclosure is determined by the appended claims.
Claims
1. A storage system, comprising: Two or more data storage devices, wherein each of the two or more data storage devices comprises a zoned namespace (ZNS) architecture; and a controller coupled to the two or more memory devices, wherein the controller is configured to: collecting thermal statistics for each of the two or more data storage devices; Analyze the collected thermal statistics; as well as Partitions are designated by selecting one or more dies within at least one of the two or more data storage devices based on the analyzed collected thermal statistics.
2. The storage system according to claim 1, wherein the controller is further configured to: receiving a partition open command; and Programs host data to the specified partition. 3 . The storage system of claim 2 , wherein receiving the partition open command occurs before analyzing the collected thermal statistics. The storage system of claim 1 , wherein collecting the thermal statistics occurs continuously. 5 . The storage system of claim 1 , wherein analyzing the collected thermal statistics occurs during an offline storage system management period.
6. The memory system of claim 1 , wherein the selected one or more dies have a first thermal statistic and the remaining one or more dies have a second thermal statistic, wherein the first thermal statistic is less than the second thermal statistic, and wherein the first thermal statistic and the second thermal statistic are a temperature distribution. 7 . The storage system of claim 1 , wherein the controller is further configured to select the one or more dies based on a predicted workload.
8. The storage system of claim 1, wherein the two or more data storage devices are arranged in rows and columns to allow airflow therebetween.
9. The storage system according to claim 8, wherein: A first data storage device of the two or more data storage devices has a first thermal statistic value, and a second data storage device of the two or more data storage devices has a second thermal statistic value; and The first thermal statistic is different from the second thermal statistic.
10. The storage system according to claim 9, wherein: The first data storage device is located at a first distance from a source of airflow, and the second data storage device is located at a second distance from the source of airflow; and The first distance and the second distance are different.
11. A storage system according to claim 8, wherein analyzing the collected thermal statistics includes analyzing the thermal statistics of a first data storage device and at least one of the thermal statistics of a second data storage device and a third data storage device, and wherein the second data storage device is arranged to be directly adjacent to the first data storage device, and the third data storage device is arranged to be diagonally adjacent to the first data storage device.
12. The storage system of claim 11, wherein the second data storage device is a plurality of data storage devices, and the third data storage device is a plurality of data storage devices.
13. A storage system according to claim 1, wherein each of the two or more data storage devices includes a plurality of dies, and wherein the controller is configured to select one or more of the plurality of dies to form the partition based on collected thermal statistics of the plurality of dies.
14. A data storage device, comprising: A memory device having a zoned namespace (ZNS) architecture, wherein the memory device comprises a plurality of dies; and a controller coupled to the memory device, wherein the controller is configured to: collecting thermal statistics for each of the plurality of dies; Analyze the collected statistics; as well as One or more dies of the plurality of dies are allocated to form a partition.
15. The data storage device of claim 14, wherein the thermal statistics of each of the plurality of dies are based at least on usage of the die, proximity of the die to an airflow source of the data storage device, and a temperature of dies adjacent to the die.
16. The data storage device of claim 14, wherein the controller is further configured to: providing the collected thermal statistics to a host device; receiving a workload based on providing the collected thermal statistics; and The access queues of the data storage device are rebalanced based on the received workload and the collected thermal statistics.
17. The data storage device of claim 14, wherein the controller is further configured to: receiving a workload type indication from a host device; receiving the thermal statistics from a thermal statistics acquisition module, wherein the thermal statistics acquisition module is disposed in the controller; and Based on the received workload type indication and the collected thermal statistics, access queues of the data storage device are rebalanced.
18. A storage system, comprising: two or more memory devices; and A controller is coupled to the two or more memory devices, wherein the controller is configured to assign a partition formed by a portion of one or more of the two or more memory devices based on a thermal statistic of each of the two or more memory devices.
19. The storage system according to claim 18, wherein the storage system is an enterprise-side storage system.
20. The storage system of claim 18, wherein the storage system is a client storage system.
Citation Information
Patent Citations
Controlled System Management Based On Storage Device Thermal Load
US20230325113A1
