International Business Machines Corporation

Other

Globalfoundries U.S. 2 LLC

Andrew Dale Walls

Roman Alexander Pletka

Charles John Camp

Nikolas Ioannou

Timothy J Fisher

Carol Spanel

Aaron Daniel Fry

Radu Ioan Stoica

Sasa Tomic

Michael T Benhase

Charalampos Pozidis

Ioannis Koltsidas

Nikolaos Papandreou

Michael Joseph Azevedo

Gary A Tressler

Cheng-Chung Song

Carl Evan Jones

Yu-Cheng Hsu

Lee Dale Cleveland

Gregg Steven Lucas

John Charles Elliott

Thomas P Parnell

William Garrett Verdoorn, Jr

James Stephen Fields, Jr

Christopher M Dennett

David Frank Hepner

Robert Akira Kubo

Brian James Cagno

Stephen LaRoux Blinick

Franck Excoffier

Antonios Kornilios Kourtis

Lih-Chung Kuo

Bitwoded Okbay

Adam C Chunn

Matthew J Kalos

Evangelos Stavros Eleftheriou

Daniel Frank Moertl

Subhojit Roy

Thomas Mittelholzer

Robert Edward Galbraith

Michael Howard Hartung

Rahul M Fiske

James Chien-Chiung Chen

Lincoln T Simmons

Michael John Palmer

Seth David Lewis

Shawn P Authement

Andrew Moy

Christopher William Mann

Hugh W McDevitt

Roger Gregory Hathorn

Karthick Rajamani

Brent Cameron Beardsley

Charles Robert Lefurgy

Robert Haas

Malcolm Scott Allen-Ware

Adalberto Guillermo Yanes

Michael O Jenkins

Kevin E Sallese

Damon W Finney

M Dean Sciacca

Donald J Lang

Jan Michael Janick

Dustin J Vanstee

Peter Paul Urbisci

Bruce Ralph Petz

David N Mora

Narasimhareddy L Annapareddy

David A Plomgren

Nhan Q Vo

Kevin A Bosien

Bryan Bordeaux Grandy

Matthew S Key

William W Hutsell

Robert W Tillerson

David S Drinnan

Kevin G Zitny

Donald J Albin

Larry B Kessler

Song C Liang

William G Verdoom, Jr

Richard P Rieger

James Chien-Chung Chen

A non-volatile memory includes physical blocks each including a respective plurality of cells, where each cell is capable of storing multiple bits of data. A controller maintains dynamically resizable pools of physical blocks, including at least a low-density pool in which cells are configured to store fewer bits and a high-density pool in which cells are configured to store more bits. The controller repeatedly dynamically resizes the low-density and the high-density pools based on write utilization of the non-volatile memory.

Storage pool resizing to meet memory endurance targets

A technique for writing data to pages in a QLC block of a QLC NAND flash memory device, where the device comprises a plurality of SLC pages organized in SLC blocks and a plurality of QLC pages organized in QLC blocks. The technique comprises storing received data in SLC pages, dividing equally a QLC block in a predefined number of sub-blocks according to a corresponding QLC page health status of the pages of the QLC block. Upon determining that SLC pages are to be copied from SLC pages to QLC pages, copying device-internal the respective SLC pages to the sub-blocks of the QLC block using device-internal cache registers, where the copying is based on an error-count aware scheme.

Write and retire pages in QLC NAND blocks

A non-volatile memory includes physical blocks each including a respective plurality of cells, where each cell is capable of storing multiple bits of data. A controller maintains dynamically resizable pools of physical blocks, including at least a low-density pool in which cells are configured to store fewer bits and a high-density pool in which cells are configured to store more bits. The controller determines whe ther total utilization of physical blocks is less than a low-density pool utilization threshold above which the high-density pool of physical blocks is utilized for storing host write data and additionally identifies cold in-use logical block address (LBA) regions having low total access frequencies. Based on determining the total utilization of physical blocks is less than the low-density pool utilization threshold, the controller performs data placement of data from the cold in-use LBA regions in blocks of the high-density pool.

Proactive data placement in high density storage by a hybrid non-volatile storage controller

A computer-implemented method, according to one embodiment, includes: maintaining a first subset of the plurality of blocks in a first pool, where the blocks maintained in the first pool are configured in SLC mode. A second subset of the plurality of blocks is maintained in a second pool, where the blocks maintained in the second pool are configured in multi-bit-per-cell mode. A current I/O rate for the memory is identified during runtime, and a determination is made as to whether the current I/O rate is outside a first range. In response to determining that the current I/O rate is not outside the first range, the blocks maintained in the first pool are used to satisfy incoming host writes. Moreover, in response to determining that the current I/O rate is outside the first range, the blocks maintained in the second pool are used to satisfy incoming host writes.

Workload based relief valve activation for hybrid controller architectures

A memory controller receives a multi-plane read request and identifies a set of actual read offsets for a set of pages in the multi-plane read request. The memory controller calculates a common read offset using the set of actual read offsets. The memory controller calculates an offset difference for. Each page. Each offset difference reflects the difference between an actual read offset for that page and the common read offset. The memory controller compares a particular page's offset difference to an offset difference threshold. The memory controller categorizes, based on the comparing, a first subset of pages from the set of pages into a single plane group and a second subset of pages from the set of pages into a multi-plane group. The memory controller performs a multi-plane read on the multi-plane group.

Triage of multi-plane read requests

A computer-implemented method, according to one approach, includes: determining a current read heat value of each logical page which corresponds to write requests that have accumulated in a destage buffer. Each of the write requests is assigned to a respective write queue based on the current read heat value of each logical page which corresponds to the write requests. Moreover, each of the write queues correspond to a different page stripe which includes physical pages, the physical pages included in each of the respective page stripes being of a same type. Other systems, methods, and computer program products are described in additional approaches.

Data placement in write cache architecture supporting read heat data separation

A computer-implemented method, according to one embodiment, includes: receiving a sub-logical page read command for data stored in NVRAM at a first LBA, and creating a searchable entry which includes the first LBA. Data read from the NVRAM is also received, where the received data corresponds to a given LBA. In response to determining that the given LBA matches the first LBA of the searchable entry, a copy of the received data is stored in a buffer. Moreover, in response to determining that a received sub-logical page write command is for data stored in the NVRAM at the first LBA, the copy of the received data in the buffer is coalesced with data included in the sub-logical page write command to form a full-logical page write. Furthermore, instructions to perform the full-logical page write in the NVRAM are sent.

Performing sub-logical page write operations in non-volatile random access memory (NVRAM) using pre-populated read-modify-write (RMW) buffers

A computer-implemented method, according to one embodiment, includes: determining a current operating state of a block of memory. The block includes more than one type of page therein, and at least one read voltage is associated with each of the page types. The current operating state of the block is further used to produce a hybrid calibration scheme for the block which identifies a first subset of the read voltages, and a second subset of the read voltages. The read voltages in the second subset are further organized in one or more groupings. A unique read voltage offset value is calculated for each of the read voltages in the first subset, and a common read voltage offset value is also calculated for each grouping of read voltages in the second subset.

Hybrid read voltage calibration in non-volatile random access memory

A computer-implemented method, according to one embodiment, includes: determining whether a number of blocks included in a first ready-to-use (RTU) queue is in a first range of the first RTU queue. In response to determining that the number of blocks included in the first RTU queue is in the first range, a determination is made as to whether a number of blocks included in a second RTU queue is in a second range of the second RTU queue. Moreover, in response to determining that the number of blocks included in the second RTU queue is not in the second range, valid data is relocated from one of the blocks in a first pool which corresponds to the first RTU queue. The block in the first pool is erased, and transferred from the first pool to the second RTU queue which corresponds to a second pool.

Adapting memory block pool sizes using hybrid controllers

A computer-implemented method, according to one embodiment, is for maintaining heat information of data while in a cache. The computer-implemented method includes: transferring data from non-volatile memory to the cache, such that the data is stored in a first page in the cache. Previous read and/or write heat information associated with the data is maintained by preserving one or more bits in a hash table which correspond to the data in the first page. Moreover, the data is destaged from the first page in the cache to the non-volatile memory, and the one or more bits in the hash table which correspond to the data are updated to reflect current read and/or write heat information associated with the data.

Increasing data read and/or write heat tracking resolution in storage devices having cache architecture

A computer-implemented method, according to one embodiment, includes: receiving, at a storage drive, a portion of a write command. Metadata information is extracted from the received portion of the write command, and sequentially added to a metadata buffer. Parity information is extracted from the received portion of the write command, and adding to a parity buffer. The data in the received portion of the write command is stored in a memory in the storage drive. A determination is also made as to whether an open segment in the memory which corresponds to the received portion of the write command has been filled. In response to determining that the open segment has been filled, the parity buffer is updated with the metadata information included in the metadata buffer. The metadata information and parity information is also destaged from the respective buffers to a physical storage location in the memory.

Metadata hardening and parity accumulation for log-structured arrays

A computer-implemented method, according to one embodiment, includes: sequentially adding metadata information that has been extracted from a received write command to a metadata buffer, and adding parity information that has been extracted from the received write command to a parity buffer. The data corresponding to the received write command is also sent to memory. A determination is made as to whether an open segment in the memory which corresponds to the write command has been filled. In response to determining that the open segment has been filled, the parity buffer is updated with the metadata information included in the metadata buffer. Moreover, the metadata information is destaged from the metadata buffer and parity information is destaged from the parity buffer to a physical storage location in the memory.

A computer-implemented method, according to one embodiment, includes: retrieving a physical block address corresponding to a logic block address, extracting information from the physical block address, and performing a lookup operation in cache using the extracted information. A range check of the physical block address is further performed in response to the lookup operation succeeding, while data is read from the cache in response to the range check succeeding. An architecture of the cache supports separation of data streams, as well as parallel writes to different non-volatile memory channels. The cache architecture further supports pipelining of the parallel writes to different non-volatile memory planes. Moreover, the non-volatile memory controller is configured to perform a direct memory lookup in the cache based on a physical block address.

Non-volatile memory controller cache architecture with support for separation of data streams

A computer-implemented method, according to one embodiment, includes: maintaining a block switching metric for each block of memory in the storage system. A determination is made as to whether a first block in a first pool should be transferred to a second pool according to a block switching metric which corresponds to the first block. In response to determining that the first block in the first pool should be transferred to the second pool according to the block switching metric which corresponds to the first block, the first block is erased. The first block is then transferred from the first pool to a second RTU queue which corresponds to the second pool. A second block in the second pool is also erased and transferred from the second pool to a first RTU queue which corresponds to the first pool.

Block mode toggling using hybrid controllers

Techniques for selecting a storage node of a storage system to store data include applying a first function to at least some data chunks of an extent to provide respective first values for each of the at least some data chunks. A storage node, included within multiple storage nodes of a storage system, is selected to store the extent based on a majority vote derived from the respective first values.

Techniques for improving deduplication efficiency in a storage system with multiple storage nodes

Growing community of inventors

San Jose, CA, United States of America

Andrew Dale Walls