Tair (Redis® OSS-Compatible):Tair minor version release notes

Minor version

Update level

Release date

Type

Description

24.12.1.0

LOW

2024-12-18

Feature optimizations

• TairVector data now supports data synchronization through DTS. The stability of data migration during scale-out and scale-in operations is improved.

Bug fixes

• Fixed an issue in TairDoc where the JSON.GET >= operator did not work.

24.12.0.0

LOW

2024-12-06

Feature optimizations

• In passwordless Sentinel mode, you can configure more passwordless commands using the #no_loose_sentinel-password-free-commands parameter.

• Improved stability.

24.11.0.0

LOW

2024-11-20

Feature optimizations

• The INFO command now returns the run_id field of the node.

• Optimized the stability of upgrades and migration for configuration changes.

• Optimized the performance of TairVector commands related to expiration.

• Optimized the stability of TairSearch and TairVector.

24.10.0.0

MEDIUM

2024-10-16

Feature optimizations

• Improved write performance by about 5% to 10%.

Bug fixes

• Fixed the CVE-2024-31228 security vulnerability.

• Fixed the CVE-2024-31227 security vulnerability.

• Fixed the CVE-2024-31449 security vulnerability.

• Improved stability.

24.8.2.0

LOW

2024-08-28

Feature optimizations

• Improved stability.

24.8.1.0

LOW

2024-08-22

Feature optimizations

• Allowed the GET parameter to be used with the NX/XX parameters in the SET command.

24.8.0.0

MEDIUM

2024-08-15

Feature optimizations

• Improved the processing speed of basic commands by 5%.

Bug fixes

• Fixed an issue where the `blocked_clients` statistic was incorrect when semi-synchronous replication was enabled. This issue only affected the statistic.

• Fixed an issue where the process could crash if the slow query queue became full.

• Optimized the stability of TairVector and TairSearch.

• Optimized the accuracy of TairRoaring memory statistics.

24.7.0.0

LOW

2024-07-22

New features

• Initial release. Compatible with Redis 7.0 and all proprietary Tair data structures.

Minor version

Update level

Release date

Type

Description

24.12.1.0

LOW

2024-12-18

Feature optimizations

• TairVector data now supports data synchronization through Data Transmission Service (DTS). Data migration is also more stable during scale-out and scale-in operations.

Bug fixes

• Fixed an issue in TairDoc where the JSON.GET >= operator did not work.

24.12.0.0

LOW

2024-12-06

Feature optimizations

• In Sentinel passwordless mode, you can configure more passwordless commands using the #no_loose_sentinel-password-free-commands parameter.

• Improved stability.

24.11.0.0

LOW

2024-11-20

Feature optimizations

• The INFO command now returns the run_id field of the node.

• Optimized the stability of upgrades and migration for configuration changes.

• Optimized the performance of TairVector commands related to expiration.

• Optimized the stability of TairSearch and TairVector.

24.10.0.0

MEDIUM

2024-10-16

Feature optimizations

• Improved write performance by about 5% to 10%.

Bug fixes

• Fixed the CVE-2024-31228 security vulnerability.

• Fixed the CVE-2024-31227 security vulnerability.

• Fixed the CVE-2024-31449 security vulnerability.

• Improved stability.

24.8.2.0

LOW

2024-08-28

Feature optimizations

• Improved stability.

24.8.1.0

LOW

2024-08-22

Feature optimizations

• Improved stability.

24.8.0.0

MEDIUM

2024-08-15

Feature optimizations

• Improved the processing speed of basic commands by 5%.

Bug fixes

• Fixed an issue where the `blocked_clients` statistic was incorrect when semi-synchronous replication was enabled. This issue only affected the statistic.

• Fixed an issue where the process could crash if the slow query queue became full.

• Optimized the stability of TairVector and TairSearch.

• Optimized the accuracy of TairRoaring memory statistics.

24.7.0.0

LOW

2024-07-22

Feature optimizations

• TairVector now supports HNSW indexes for sparse vectors.

• Added the ts-auto-del-empty-skey-enable parameter to control the expiration behavior of Skeys in TairTS. The default value is yes, which means an Skey is deleted when all its data points expire.

• Improved stability.

24.6.1.1

LOW

2024-07-09

Feature optimizations

• Improved the stability of TairSearch.

24.6.1.0

LOW

2024-07-05

New features

• TairDoc now supports the JSON.MERGE command.

Feature optimizations

• Improved the stability of TairVector.

24.6.0.0

LOW

2024-07-02

Feature optimizations

• Optimized synchronization stability between primary and secondary nodes in high-traffic burst scenarios.

24.5.1.0

MEDIUM

2024-06-06

Feature optimizations

• Removed the usage restrictions on the COMMAND GETKEYS command.

• Optimized the memory usage of FLAT vector indexes in TairVector to reduce memory waste for small indexes.

• Optimized inverted indexes such as Within and multi-condition Condition in TairVector, and added the TVS.KNNSEARCHFIELD and TVS.MINDEXKNNSEARCHFIELD commands.

• Added the field_type field to the aux_info returned when the TFT.MSEARCH command in TairSearch sorts by an indexed field.

Bug fixes

• Fixed a write performance regression in TairVector that began in version 24.3.2.2. Upgrade to this version or later.

• Fixed several issues in TairSearch and optimized its stability.

24.5.0.0

MEDIUM

2024-05-22

Feature optimizations

• Optimized the synchronization capability of cross-slot commands in active-passive mode.

• Optimized the speed of background BGREWRITE for large instances.

• The INFO command now returns CPU, QPS, and traffic information for each command.

Bug fixes

• Improved stability.

24.4.1.0

MEDIUM

2024-04-22

Feature optimizations

• Supports arbitrary rollback of synchronization points after a DTS task restarts. If your instance requires a long-term DTS task configuration, upgrade to this version.

24.4.0.0

MEDIUM

2024-04-16

Feature optimizations

• Optimized the memory usage of basic data structures to reduce memory usage for a small number of complex structure elements.

• Optimized the retrieval performance of TairVector's combined full-text and scalar retrieval.

• Similar to single-index retrieval in TairVector, multi-index retrieval now supports returning non-vector fields.

Bug fixes

• Improved stability.

24.3.2.2

MEDIUM

2024-03-21

Feature optimizations

• Optimized the engine metadata usage for standard architecture (primary/secondary) instances.

Bug fixes

• Improved stability.

24.3.2.1

MEDIUM

2024-03-18

Feature optimizations

• Enhanced the scalar-first, then vector retrieval feature of TairVector.

• Optimized the TairVector scalar index storage mechanism to improve performance with duplicate index values. The maximum index value length is now 32 KB.

• TairVector now supports returning query results from pure scalar inverted indexes.

Bug fixes

• Fixed an issue where the attribute_data_size and index_data_size item statistics in the TVS.GETINDEX interface of TairVector were inaccurate.

• Fixed an issue where the format returned by the XINFO command had an extra level of indentation.

• Fixed an issue where the number of results returned by the SCAN command could be much larger than the Count parameter value.

24.3.1.0

LOW

2024-03-11

Feature optimizations

• Delayed the creation of metadata for multiple databases (DBs). Metadata is now created when data is read from or written to a DB, instead of immediately upon executing the SELECT command. This prevents metadata bloat caused by some Redis management tools that traverse all DBs.

• Optimized the expiration check efficiency when multiple DBs exist, fairly dispatching expiration tasks based on the amount of data with expiration times.

24.3.0.0

MEDIUM

2024-03-07

Feature optimizations

• Added support for second-level rollback in case of a smooth scaling migration failure, preventing instance unavailability due to a failed configuration change.

  Note

  However, using many logical DBs can lead to high metadata memory usage. If you encounter this issue, upgrade to version 24.3.2.2 or later, where this issue has been optimized.

• Allowed the execution of multi-DB commands SWAPDB, COPY, and MOVE in a cluster architecture.

Bug fixes

• Fixed an out-of-memory (OOM) issue that could occur when the LIMIT for the ZRANGEBYSCORE command was set too high, causing the DB to pre-allocate excessive memory.

24.1.0.0

MEDIUM

2024-01-23

Feature optimizations

• TairVector supports creating inverted indexes for scalars (such as labels and attributes) and enables scalar-first, then vector KNN retrieval.

Bug fixes

• Fixed an issue where the GETDEL command could cause an instance to crash under abnormal conditions.

• Fixed an issue where the last-entry information returned by the XINFO STREAM command was incorrect.

• Fixed several issues in TairVector and optimized its stability.

23.12.2.0

LOW

2023-12-26

New features

• Added support for semi-synchronous mode (semisync), which can be controlled using the #no_loose_tsync-repl-mode parameter. This is a proof of concept (POC) version. Use it with caution in production environments.

23.12.1.2

MEDIUM

2023-12-21

Feature optimizations

• Optimized write performance in some scenarios. An upgrade is recommended.

• TairSearch now supports the EXPAINSCORE command.

Bug fixes

• Merged multiple CVE fixes for Lua scripts from the Redis community.

• Fixed several issues in TairSearch and TairVector.

• Optimized several stability issues. An upgrade is recommended.

23.8.1.2

MEDIUM

2023-08-22

Feature optimizations

• Reduced CPU usage when the instance is idle and improved basic instance performance.

• Standardized the CPU usage metric in CloudDBA.

• Primary-secondary replication traffic is now displayed separately in INFO STATS.

Bug fixes

• Fixed several issues in TairVector.

23.8.0.0

MEDIUM

2023-08-03

New features

• Full support for TLS encrypted connections.

• TairVector now supports setting a TTL at the key level within an index, supports vector nearest neighbor search for a specified list of keys, and supports full-text search, enabling hybrid search that combines vector search and full-text search.

Feature optimizations

• Optimized overall performance and improved stability.

• TairZset now supports the EXZRANKBYSCORE command.

• The behavior of TairDoc interfaces is now aligned with the current TairDoc official documentation.

• Optimized the performance of commands such as GETBIT, BITPOS, and BITCOUNT.

Bug fixes

• Fixed several issues in TairSearch and TairVector.

6.2.8.4

MEDIUM

2023-07-13

Bug fixes

• Fixed a rare issue where a Filter in TairVector could cause an instance to crash.

6.2.8.3

MEDIUM

2023-07-12

Feature optimizations

• Optimized the filter limits in TairVector to shorten running time.

6.2.8.2

MEDIUM

2023-07-04

Feature optimizations

• Optimized the stability of automatic garbage collection for TairVector HNSW indexes.

6.2.8.1

MEDIUM

2023-07-02

Bug fixes

• Fixed the following issue: After the RENAME command is executed on the primary node and synchronized to the secondary node, the secondary node crashes if the destination key has expired.

6.2.8.0

MEDIUM

2023-07-01

New features

• TairVector now supports automatic garbage collection for HNSW indexes.

Bug fixes

• Fixed an issue where using the WAIT command in a cluster architecture could cause a crash.

• Fixed an issue where executing EVALSHA from different clients after a SCRIPT LOAD command could occasionally report a NOSCRIPT error.

6.2.7.3

MEDIUM

2023-06-12

Feature optimizations

• Optimized background backup performance and increased backup speed.

6.2.7.2

MEDIUM

2023-06-12

New features

• TairSearch now supports array-type documents and the Okapi BM25 similarity algorithm.

Feature optimizations

• Optimized the memory usage of the Dict structure.

Bug fixes

• Fixed a matching issue with the Pattern Tokenizer in TairSearch and improved stability.

6.2.6.9

LOW

2023-05-09

Feature optimizations

• Improved stability.

6.2.6.8

MEDIUM

2023-04-28

Bug fixes

• Fixed an issue where horizontal scaling tasks might not finish correctly under certain conditions.

6.2.6.7

LOW

2023-04-22

Feature optimizations

• Allows user accounts to execute the ROLE command.

6.2.6.6

LOW

2023-04-20

Feature optimizations

• Improved the stability of TairSearch.

6.2.6.5

MEDIUM

2023-04-17

Feature optimizations

• Optimized the logic and stability of TairSearch.

Bug fixes

• Fixed an issue where the laddr address displayed in the client list could be incorrect.

6.2.6.2

HIGH

2023-04-12

Security hardening

• Merged all recent critical CVE security fixes from the Redis community to improve stability and security.

6.2.5.3

LOW

2023-03-20

Bug fixes

• Fixed a storage exception issue in TairSearch.

6.2.5.2

LOW

2023-03-17

Feature optimizations

• Optimized the primary-secondary synchronization logic for TairSearch in extreme scenarios.

Bug fixes

• Fixed an issue where TairSearch did not support integers when parsing fields of type Double.

6.2.5.0

HIGH

2023-03-14

Feature optimizations

• Optimized the TairString data structure, saving 8 bytes of memory per key.

• TairSearch now supports compressed document storage and adds the TFT.ANALYZER command to view tokenizer effects.

Bug fixes

• Fixed some issues with memory statistics in TairSearch.

• Fixed an issue where clients blocked by Blocking-related interfaces might not be woken up.

• Fixed an incompatibility issue with the results returned by the XINFO command in some cases.

6.2.4.7

HIGH

2023-03-03

Bug fixes

• Fixed an issue where a high volume of Pipeline requests could cause High Availability (HA) problems in extreme cases.

6.2.4.6

LOW

2023-03-02

Feature optimizations

• The read/write splitting architecture supports executing read-only requests, such as EVAL and EVALSHA, on secondary nodes. Use the readonly_lua_route_ronode_enable parameter to control this behavior.

6.2.4.3

HIGH

2023-02-14

Feature optimizations

• Updated TairString to the latest encoding format to save more memory.

Bug fixes

• Fixed an issue where data synchronization could occasionally fail to complete when adding or removing shards in a cluster architecture.

6.2.4.1

LOW

2023-02-07

New features

• Added support for the TairSearch data structure.

• The JSON.GET command in the TairDoc data structure now fully supports JSONPath syntax.

6.2.3.2

HIGH

2023-01-13

Bug fixes

• Fixed an issue where expired data was cleared more slowly after adding or removing shards from a cluster instance.

6.2.3.1

LOW

2023-01-04

Feature optimizations

• Optimized the HA stability of read/write splitting instances.

6.2.3.0

LOW

2022-12-26

Feature optimizations

• TairVector now supports cluster proxy mode, adds the FLOAT16 vector data type, and adds multi-vector index queries.

6.2.2.0

LOW

2022-11-22

Feature optimizations

• TairVector now supports the Jaccard distance function.

• TairVector now supports statistics on the memory usage of each index (index_data_size and attribute_data_size).

6.2.1.5

LOW

2022-11-14

Feature optimizations

• Fixed an issue with the Blocking interface and improved stability.

6.2.1.3

LOW

2022-10-28

Feature optimizations

• Improved the stability of TairVector.

6.2.1.2

LOW

2022-10-14

Feature optimizations

• Improved the stability of TairVector.

6.2.1.1

LOW

2022-10-13

Initial release

• Initial release. Compatible with open source Redis 6.2 and Tair's proprietary data structures (TairSearch is not yet supported).

• Supports slow query detection and isolation for commands such as KEYS, SMEMBERS, HGETALL, and EXHGETALL.

• Further optimized performance. Compared with Redis Open-Source Edition instances of the same specifications, the performance of all interfaces, such as PUB/SUB and Lua commands, is more than doubled.

• Added the proprietary TairVector for vector search. It supports HNSW and FLAT index algorithms and provides a high-performance, real-time vector database service that integrates storage and retrieval.

Minor version

Update level

Release date

Type

Description

5.0.57

MEDIUM

2024-12-10

Feature optimizations

• Optimized cumulative traffic and command statistics for clients.

Bug fixes

• Fixed an issue where the large key statistics threshold became invalid during cluster instance scaling.

• Improved the stability of cluster instance scaling.

5.0.56

LOW

2024-11-15

Bug fixes

• Fixed issues related to the time-to-live (TTL) in the TairString module.

• Fixed issues in the TairTS module.

5.0.55

MEDIUM

2024-10-08

Feature optimizations

• Added support for QPS throttling.

Bug fixes

• Fixed multiple issues in the TairSearch module.

• Fixed the CVE-2024-31228 security vulnerability.

• Fixed the CVE-2024-31449 security vulnerability.

5.0.54

LOW

2024-08-28

Feature optimizations

• Compatible with authentication using an incorrect password when the VPC passwordless feature is enabled.

5.0.53

LOW

2024-07-23

Feature optimizations

• Optimized the whitelist check logic for when both the VPC passwordless feature and the #no_loose_check-whitelist-always parameter are enabled.

• The latency insight feature now supports the monitoring of Pipelines.

• Compatible with open source Redis 5.0.14.

Bug fixes

• Fixed an issue where the execution duration of commands in Lua and Multi was calculated incorrectly. This bug affected versions 5.0.49 to 5.0.52.

5.0.52

LOW

2024-06-25

Feature optimizations

• Optimized the execution efficiency of Pub/Sub commands to prevent stuttering when many connections unsubscribe at the same time.

• Optimized the client polling check policy to prevent memory statistics and collection from being delayed for persistent connections.

• Optimized the disk writing method for child processes that generate RDB and AOF files.

5.0.51

LOW

2024-05-28

Feature optimizations

• Improved stability.

5.0.50

LOW

2024-05-28

Feature optimizations

• Added monitoring for connections in the WATCH or BLOCK state.

• Added monitoring metrics related to Rehash.

• Upgraded the real-time hot key statistics feature. It now supports displaying precise QPS and can track up to 50 hot keys at the same time.

Bug fixes

• Fixed an issue with incorrect QPS statistics caused by shared objects. Shared objects are no longer used when the real-time hot key statistics feature is enabled.

5.0.49

MEDIUM

2024-04-24

Feature optimizations

• The INFO STATS command now returns statistics on disconnections due to exceeding client input and output buffer limits:

  • client_query_buffer_limit_disconnections

  • client_output_buffer_limit_disconnections

• A real-time threshold for large key statistics is now available. The default value is 2,000. For example, a String type is considered a large key if its character length exceeds 2,000. List, Set, and Hash types are considered large keys if their element count exceeds 2,000.

Bug fixes

• Fixed a crash issue in the EXZRANKBYSCORE and EXZREVRANKBYSCORE commands of exZset by adding a check for the Score dimension.

5.0.48

LOW

2024-01-09

Feature optimizations

• Optimized the efficiency of active expiration.

• Added the pubsub_clients metric to the INFO CLIENTS command.

• TairSearch now supports the EXPAINSCORE command.

• Improved the stability of TairSearch.

Bug fixes

• Fixed a memory bloat issue in the TairSearch tokenizer.

5.0.47

LOW

2023-10-18

Feature optimizations

• Fixed an issue where CPU usage could reach 100% in extreme scenarios.

5.0.46

LOW

2023-10-10

Feature optimizations

• Optimized the connection type (TCP or TLS) detection for Lua scripts in a cluster architecture.

5.0.45

MEDIUM

2023-09-20

Feature optimizations

• Optimized the performance of the CLUSTER SLOTS and CLUSTER NODES commands.

Bug fixes

• Fixed an issue where TairHash could occasionally crash.

5.0.44

MEDIUM

2023-08-24

Feature optimizations

• Added support for splitting traffic into data traffic and replication traffic.

Bug fixes

• Fixed the CVE-2022-24834 security vulnerability.

5.0.41

LOW

2023-07-21

Feature optimizations

• Improved the stability of TairSearch.

5.0.35

LOW

2023-06-12

New features

• TairSearch now supports the Okapi BM25 similarity algorithm.

5.0.34

LOW

2023-05-22

Feature optimizations

• Improved the stability of smooth scaling.

5.0.33

LOW

2023-04-23

New features

• TairSearch now supports array-type documents.

• Improved the stability of TairSearch.

5.0.32

LOW

2023-03-31

Feature optimizations

• TairSearch adds the TFT.EXPLAINCOST command.

5.0.30

LOW

2023-03-20

Feature optimizations

• Improved stability.

5.0.29

LOW

2023-03-16

Feature optimizations

• Improved stability.

5.0.28

LOW

2023-03-15

Feature optimizations

• TairSearch now supports compressed document storage and adds the TFT.ANALYZER command to view tokenizer effects.

5.0.27

HIGH

2023-02-15

Feature optimizations

• Periodically prints hot key information to the audit log for easy viewing.

Bug fixes

• Fixed a rare crash that could occur when scaling a cluster instance after Global Distributed Cache is enabled.

5.0.25

LOW

2023-01-13

Feature optimizations

• TairSearch fully supports the tokenizer framework. For more information, see Search tokenizers.

• The term and terms queries in TairSearch support the lowercase parameter.

5.0.22

MEDIUM

2022-11-02

Feature optimizations

• TairSearch now supports IK and Pattern tokenizers.

• Improved stability.

5.0.21

LOW

2022-11-02

Feature optimizations

• Improved stability.

5.0.18

LOW

2022-09-29

Feature optimizations

• Improved the stability of TairSearch data parsing and persistence.

• TairSearch now supports Pipelines for Query Builder clients.

5.0.15

LOW

2022-08-29

Feature optimizations

• Published the MSEARCH feature for TairSearch.

• Enhanced the stability of multi-threaded writes in TairSearch.

• Improved the stability of cluster architecture instances.

5.0.10

LOW

2022-07-28

Feature optimizations

• Improved stability.

5.0.9

LOW

2022-07-27

Feature optimizations

• Enhances the stability of scale-out and scale-in operations for Global Distributed Cache instances.

• Improved the accuracy of Sort (result sorting) in TairSearch.

• Improved the query efficiency of TairSearch in sparse document scenarios.

• Account names are now case-sensitive.

• Removed the query optimization for the KEYS command to avoid unexpected behavior.

1.9.11

MEDIUM

2024-01-03

Bug fixes

• Fixed an issue where TairTS was missing a return value in some cases.

1.9.9

MEDIUM

2023-03-23

Bug fixes

• Fixed an issue where the return value type of the TairRoaring TR.GETBIT command was incorrect.

1.9.8

LOW

2023-03-16

Feature optimizations

• Improved stability.

1.9.7

LOW

2022-12-26

Feature optimizations

• Improved the stability of TairSearch.

1.9.6

LOW

2022-11-02

Feature optimizations

• Improved the stability of TairTS.

1.9.4

LOW

2022-08-24

Feature optimizations

• Optimized the startup process for cluster architecture instances.

1.8.9

LOW

2022-06-07

Feature optimizations

• Improved the service stability of instances after direct connection access is enabled.

1.8.8

LOW

2022-06-01

Feature optimizations

• Updated TairSearch to improve the efficiency of combined queries.

1.8.5

HIGH

2022-05-24

Security hardening

• Improved the stability of configuration changes for cluster instances after direct connection access is enabled.

• Added support for the TairSearch aggregation feature and improved document writing efficiency.

1.8.4

LOW

2022-05-17

Feature optimizations

• Improved the stability and execution efficiency of TairSearch.

1.8.3

LOW

2022-04-28

Feature optimizations

• Improved the stability of TairTS.

1.8.2

HIGH

2022-04-22

Security hardening

• Fixed a rare issue where migrating large keys failed during a configuration change for a cluster instance with direct connection access enabled.

Feature optimizations

• Optimized TairSearch query efficiency.

1.8.1

LOW

2022-04-20

Feature optimizations

• Released TairRoaring V2.2.

1.7.28

LOW

2022-03-24

Feature optimizations

• The latency histogram now supports statistics for Tair module commands.

1.7.27

LOW

2022-03-11

Feature optimizations

• Released TairRoaring V2.

• Released TairSearch.

1.7.20

MEDIUM

2022-01-04

Feature optimizations

• Optimized TairRoaring performance.

1.7.17

LOW

2021-11-29

Feature optimizations

• Fixed an issue where a cluster instance failed to restart under abnormal conditions.

1.7.16

LOW

2021-11-24

Feature optimizations

• Fixed an issue where the real-time key analysis feature missed statistics for the Spop command.

1.7.12

MEDIUM

2021-10-26

Feature optimizations

• Fixed an issue of excessive slow query logs during cluster configuration changes and improved stability.

1.7.11

MEDIUM

2021-10-15

Feature optimizations

• Added transparent data encryption (TDE) information to the Info result.

• Improved stability.

1.7.9

LOW

2021-10-13

Feature optimizations

• Enhanced the autonomous capability of seamless slot migration during cluster upgrades or downgrades.

1.7.8

MEDIUM

2021-09-26

Feature optimizations

• Optimized the Raycast algorithm in the TairGis data structure to fix an issue where the GIS.CONTAINS command returned inaccurate search results in some cases.

1.7.7

MEDIUM

2021-09-13

Feature optimizations

• Released support for TairRoaring.

• Improved stability.

1.7.6

MEDIUM

2021-08-30

Feature optimizations

• Fixed a memory leak during an append-only file (AOF) persistence rewrite.

• Subdivided Queries Per Second (QPS) statistics to support read, write, and other categories.

1.7.5

MEDIUM

2021-08-16

Feature optimizations

• Improved the reliability of smooth slot migration and enhanced stability.

1.7.4

HIGH

2021-08-11

Bug fixes

• Fixed a rare issue where a configuration change could fail for a cluster instance with direct connection access enabled.

1.7.2

MEDIUM

2021-07-27

Feature optimizations

• Improved stability.

1.7.1

MEDIUM

2021-07-20

New features

• Added support for Transparent Data Encryption (TDE), which encrypts and decrypts RDB data files.

Feature optimizations

• Optimized the memory usage of the TairHash data structure.

1.6.15

MEDIUM

2021-07-12

Feature optimizations

• Improved stability.

1.6.9

LOW

2021-06-22

New features

• Added support for the latency histogram.

1.6.6

MEDIUM

2021-06-08

Feature optimizations

• Improved stability.

1.6.3

LOW

2021-05-17

New features

• Supports retaining some keys by key or key pattern when clearing data. This can be used for:

  • Restoring some keys or key patterns when performing a point-in-time restore using data flashback.

  • Deleting or retaining keyspace content by key or key pattern.

1.6.2

MEDIUM

2021-04-25

Feature optimizations

• Optimized performance and migration speed in smooth scaling scenarios.

• Added support for obtaining a virtual IP address (VIP) over the public network to provide better support for clients using the direct connection mode.

• Optimized the large key format.

1.6.1

MEDIUM

2021-04-08

New features

• Enabled the big key statistics feature by default.

• TairString now supports `gt version`.

Feature optimizations

• Enhanced capability for slot migration and seamless scale-out.

1.5.1

HIGH

2021-03-15

Bug fixes

• Fixed a correctness issue in real-time big key statistics when overwriting a key with the same name.

1.5.0

MEDIUM

2021-02-22

New features

• Added support for real-time big key statistics.

Feature optimizations

• Optimized the liveness detection capability of the high availability (HA) system when calling fork in large memory scenarios to avoid potential long pauses.

1.4.16

HIGH

2021-01-11

New features

• When a multi-active link built with Global Distributed Cache or DTS is released, metadata can be cleared quickly to enable a fast next sync.

Bug fixes

• Fixed a memory leak issue caused by frequently executing FLUSHALL in TairHash scenarios.

1.4.13

LOW

2020-11-27

New features

• When the illegal address error message is returned, Tair includes the current client's IP address in the error message. Set the correct IP whitelist for the Tair instance based on the prompt.

  Figure 1. IP address prompt

1.4.12

MEDIUM

2020-11-26

New features

• TairString extends its syntax with flags, such as FLAGS, to be compatible with Memcache semantics.

Feature optimizations

• Enhanced the liveness detection capability of the high availability (HA) system.

  Note

  Upgrade to minor version 1.5.0 or later to get the latest optimizations for this feature.

1.4.9

HIGH

2020-10-22

Bug fixes

• Fixed an issue where TairString expiration generated incorrect binary logs, which could cause data inconsistency between primary and secondary instances.

• Fixed an issue where TairHash still performed Active Expire in read-only scenarios, causing abnormal High Availability (HA) switchovers.

• Fixed a crash that could occur when restarting an instance after a forced stop.

• Fixed an issue where expired keys in an RDB file were deleted when the file was loaded during a point-in-time restore using data flashback.

1.4.8

HIGH

2020-10-14

Bug fixes

• Fixed a memory leak issue that could occur when loading some modules.

1.4.7

MEDIUM

2020-10-12

Feature optimizations

• Caches the execution result of the CLUSTER NODES command to optimize its execution speed.

1.4.6

MEDIUM

2020-09-28

Feature optimizations

• Optimized the processing capability of some modules in special scenarios.

1.4.1

MEDIUM

2020-09-08

New features

• Proxy nodes now support passing through client IP addresses to operational logs and audit logs, making it easier to interpret logs and locate specific clients.

Feature optimizations

• Optimized data ingestion to avoid impacting data sharding when there are too many connections and I/O threads are busy.

1.3.17

MEDIUM

2020-08-04

Feature optimizations

• Optimized link latency for two-way data synchronization scenarios using DTS.

1.3.16

HIGH

2020-07-19

New features

• Added support for the ECS security group feature. You can quickly grant authorization by attaching the security group of an ECS instance to a Tair instance, without manually entering the ECS IP address. This simplifies O&M.

• Updated the TairString module to support more APIs (flags) compatible with Memcache semantics.

Bug fixes

• Fixed an issue where BGREWRITEAOF was interrupted during a point-in-time restore using data flashback.

• Corrected the latency mark in audit logs to avoid confusion in primary and secondary audit logs.

1.3.9

MEDIUM

2020-06-19

Feature optimizations

• When using Global Distributed Cache or DTS for multi-way data synchronization, clearing data now automatically cleans up saved point-in-time metadata. This ensures a fast recovery for the replicator.

1.3.7

LOW

2020-05-19

New features

• In the return value of the INFO command, the Replication section now includes role information (such as role:master). This is compatible with the Redisson client's call for this information in some scenarios.

1.3.6

MEDIUM

2020-05-19

New features

• Periodically prints hot key information to the log for easy viewing.

• The performance statistics feature now distinguishes QPS for read, write, and sync operations, making statistics more accurate.

Feature optimizations

• Optimized the kernel capability for point-in-time restore using data flashback, simplifying the data restoration process.

• Commands such as AUTH, ADMINAUTH, and CONFIG no longer record sensitive information, improving security.

1.3.5

HIGH

2020-04-22

Bug fixes

• Fixed a deadlock issue that could occur in the multi-threaded engine when asynchronously releasing client connections.

• Fixed an issue where the file descriptor (FD) in the engine could not be linearly expanded.

1.3.3

HIGH

2020-04-22

New features

• TairBloom switched to a 64-bit hash algorithm.

• TairBloom now adds an estimation of final memory usage for accurate memory statistics.

• TairHash adds the exhgetAll2 interface to correct the command's response format.

Bug fixes

• Corrected the error message returned by the instance when the whitelist is not set correctly, from (error) ERR invalid password to (error) ERR illegal address.

• Fixed a memory leak issue that could occur when operating on multiple POLYGONs with TairGIS.

• Fixed the default path issue in TairDoc.

• Fixed a race condition issue that could occur with Pub and Sub commands in the multi-threaded engine.

1.3.1

HIGH

2020-04-03

New features

• The data flashback feature restores Tair data to any point in time within the last 7 days. This feature prevents data loss from accidental operations, significantly reduces O&M complexity, and provides real-time data protection.

• TairGIS is compatible with Redis GEO-related commands.

• TairBloom supports capacity safety checks for the BFRESERVE interface.

• TairHash supports new features:

  • Commands such as EXHSET, EXHEXPIRE, EXHINCRBY, and EXHINCRBYFLOAT now include the NOACTIVE option, which reduces memory overhead in some scenarios.

  • EXHINCRBY now includes the MAX and MIN options to enforce boundaries.

  • EXHLEN now includes the noexp option to return the actual length.

• The Hash data structure supports the HINCRBY and HINCRBYFLOAT commands. Use the transactional capability of these commands to simultaneously increment or decrement multiple fields for the same key.

Feature optimizations

• Optimized the use of data structure modules.

• Significantly improved the performance of MGET and MSET for JedisCluster clients in cluster instances.

Bug fixes

• Fixed a binary log space overflow issue.

• Fixed a crash issue that could occur when a hot spot key was evicted.

• Fixed a crash issue in TairHash caused by a double deallocation.

• Fixed a crash issue caused by Use-After-Free (UAF) when disabling audit logs.

1.0.10

LOW

2020-02-19

New features

• Added the BITFIELD_RO command, which significantly optimizes performance in read/write splitting scenarios.

  Note

  If the BITFIELD command only has the get option, the proxy node converts this command to BITFIELD_RO and forwards it to multiple backend data shards.

1.0.9

HIGH

2020-02-19

Bug fixes

• Fixed a crash issue in the replication process caused by executing some complex commands in a Lua script.

1.0.8

HIGH

2020-02-10

Feature optimizations

• Optimized the algorithm and performance of traffic throttling.

Bug fixes

• Fixed an issue where a client output buffer backlog could trigger server-side overload protection, leading to a service crash.

1.0.5

HIGH

2020-02-01

New features

• Added support for the Global Distributed Cache feature. This is a proprietary multi-active database system from Alibaba Cloud. It supports business scenarios where multiple sites in different regions provide services at the same time. It helps enterprises quickly replicate Alibaba's active geo-redundancy architecture.

• Added the binary log (binlog) feature and its protocol to support services such as Global Distributed Cache.

Bug fixes

• Fixed an issue to ensure the INFO command response includes cluster_enabled information in direct connection mode. This allows some SDKs to correctly self-negotiate to cluster mode.

• Fixed inaccurate statistics for the number of management clients.

• Fixed a crash that could occur when a client is released.

• Fixed a crash that could occur when a pipeline contains complex commands.

0.2.9

HIGH

2020-01-06

Feature optimizations

• Optimized the memory usage of TairHash.

Bug fixes

• Fixed a crash issue that could occur during traffic throttling.

0.2.7

HIGH

2019-12-23

New features

• Added support for direct connection mode. Clients can bypass the proxy using a direct connection address, similar to connecting to a native Redis cluster. This reduces link overhead and further improves the instance's response speed.

• Added support for hot spot key detection logic and optimized memory usage to accurately discover the engine's hot spot keys.

• Added support for the EXCAS command, which is suitable for optimistic locking scenarios.

Bug fixes

• Fixed a coredump issue that could occur when using a pipeline.

0.2.3

LOW

2019-12-03

Initial release

Tair (Enterprise Edition) memory-optimized is suitable for business scenarios with high concurrency, many read/write hot spots, and extremely high performance requirements. Compared to Redis Open-Source Edition, its advantages are as follows:

• It uses a multi-threaded model, and its performance is about three times that of a Redis Open-Source Edition instance with the same specifications.

• It provides multiple enhanced data structure modules, including TairString (with CAS and CAD), TairHash, TairGIS, TairBloom, and TairDoc. These help you streamline a large amount of code, improve overall business performance, and focus on business innovation.

Minor version

Update level

Release date

Type

Description

2.5.4

LOW

2024-11-05

New features

• SSD (local disk) instances support data flashback.

2.5.3

HIGH

2024-07-05

New features

• Added support for GEO series commands.

• Added support for the RENAME and RENAMENX commands, but they can only modify keys up to the size of max-rename-commit-size (default is 16 MB).

• Added support for the ZPOPMIN, ZPOPMAX, BZPOPMIN, and BZPOPMAX commands.

• Added support for the LPOS command.

• Optimized the encoding of the List data structure to better match the CPU complexity of the memory-optimized List data structure.

Bug fixes

• Fixed the String Rewrite interface so that earlier versions of Redis can recognize it. When you use DTS as a source database for migration, the destination can be an earlier version of a Redis instance.

• Fixed the Redis version displayed in the INFO command from 4.0 to 6.0.

• Improved stability.

2.5.2

HIGH

2024-01-22

New features

• Supports use as a source database for DTS migration and synchronization, and supports resumable transfers.

• After a client disconnects, Tair terminates commands that can cause slow queries, such as SUNION, SUNIONSTORE, SINTER, SINTERCARD, SINTERSTORE, SDIFF, SDIFFSTORE, ZUNIONSTORE, and ZUNIONSTORE, to reduce resource consumption. Write commands might not execute successfully at this time.

Feature optimizations

• Optimized the lock granularity of background garbage collection.

Bug fixes

• Improved the stability of version upgrades when the disk is at a high watermark.

• Fixed a bug where data statistics could be overestimated.

• Improved the stability of some commands.

• Improved overall instance stability.

2.4.2.5

HIGH

2023-12-20

Bug fixes

• Improved stability.

2.4.2.4

HIGH

2023-11-06

Bug fixes

• Improved the stability of garbage collection.

2.4.2.3

HIGH

2023-11-01

Feature optimizations

• Background garbage collection deletes large keys in small batches. The default batch size is 1 MB.

2.4.2.2

HIGH

2023-09-04

Bug fixes

• Improved the stability of the FLUSHALL command and the backup interface.

Feature optimizations

• Improved the speed of FLUSHDB.

• Added logic to limit the speed of background writes to disk.

2.4.2.1

HIGH

2023-08-23

Feature optimizations

• Improved the stability of iteratively reading Hash, Set, List, and ZSet data structures when a large amount of garbage data exists.

2.4.1.3

HIGH

2023-07-14

Bug fixes

• Fixed an issue where the SSCAN command returned an exception in specific scenarios.

2.4.1.2

HIGH

2023-07-11

Bug fixes

• Improved stability.

2.4.1.1

HIGH

2023-07-10

Bug fixes

• Improved stability.

2.4.1

HIGH

2023-07-04

Feature optimizations

• Optimized the read and write performance of instances in some extreme scenarios.

• Optimized log space usage.

2.3.3

HIGH

2023-04-20

Bug fixes

• Fixed an issue where clients blocked by Blocking-related commands might not be woken up.

2.3.2

HIGH

2022-12-22

Bug fixes

• Fixed a rare issue where commands such as BRPOP, BLPOP, and BRPOPLPUSH could cause a crash.

2.3.1

MEDIUM

2022-09-14

New features

• Supports transactions (MULTI and EXEC commands), controlled by the txn-isolation-lock parameter.

• Supports Lua scripts (EVAL, EVALSHA, and SCRIPT commands), controlled by the txn-isolation-lock and #no_loose_lua-strict-mode parameters.

• Supports String Bitmap-related commands (BITCOUNT, BITFIELD, BITOP, BITPOS, GETBIT, and SETBIT).

Feature optimizations

• Optimized the stability of primary-secondary synchronization.

2.2.22

MEDIUM

2022-08-10

Feature optimizations

• Optimized the continuous read and write performance for large-capacity instances over 1 TB.

2.2.15

HIGH

2022-06-06

Bug fixes

• Improved stability.

2.1.13

LOW

2022-05-05

Feature optimizations

• Optimized the stability of primary-secondary synchronization.

2.1.12

LOW

2022-04-26

Feature optimizations

• Optimized the stability of primary-secondary synchronization.

2.1.10

LOW

2022-04-14

Feature optimizations

• Optimized primary-secondary synchronization efficiency for greater throughput and lower latency during synchronization.

• Optimized the resource consumption of Compact tasks generated when many large keys expire at the same time.

2.1.7

LOW

2021-08-12

Feature optimizations

• Optimized the speed of primary-secondary switchovers (active HA) and improved stability.

2.1.5

HIGH

2021-07-13

Bug fixes

• Fixed an issue where binary log files used too much space.

2.1.4

MEDIUM

2021-07-07

Feature optimizations

• Optimized storage parameters for TB-level capacity instances to improve performance.

• Improved stability.

2.1.0

LOW

2021-05-26

New features

• Phased release of the transaction feature (supports MULTI and EXEC commands).

• Phased release of the Lua script feature (supports EVAL, EVALSHA, and SCRIPT commands).

2.0.13

LOW

2021-04-24

New features

• Added support for Keyspace Event notifications.

1.2.17

HIGH

2021-02-04

Bug fixes

• Fixed an issue where FLUSHALL could cause data inconsistency between primary and standby nodes in extreme cases. This change enhances data consistency protection.

1.2.16

HIGH

2021-01-29

Bug fixes

• Fixed an issue where the primary/standby connection failed to automatically reconnect after a disconnection in extreme cases, which improves data consistency protection.

1.2.15

MEDIUM

2021-01-20

New features

• Added the cmd_slowlog_count parameter to track the cumulative count of slow logs.

Feature optimizations

• Optimized mitigation capabilities for when the disk is full.

1.2.14

LOW

2020-12-31

New features

• Added primary-secondary data validation.

1.2.13

HIGH

2020-12-31

Bug fixes

• Fixed a bug in the RESTORE command in some scenarios.

1.2.12

MEDIUM

2020-12-23

Feature optimizations

• Optimized the execution mechanism of the FLUSHALL and FLUSHDB commands to restrict them to sequential execution.

1.2.11

MEDIUM

2020-12-10

Feature optimizations

• Optimized memory management.

1.2.10

LOW

2020-12-04

New features

• Added support for passwordless access in a virtual private cloud (VPC).

• Added support for limiting the size of a single transaction with the max-write-batch-size option to prevent memory overflow from oversized transactions.

1.2.9

MEDIUM

2020-11-27

Feature optimizations

• Added the redis_version parameter to the response of the INFO command.

1.2.8

HIGH

2020-11-20

Bug fixes

• Fixed an issue of incorrect connection count statistics.

• Fixed an issue where the RESTORE command incorrectly counted keys.

1.2.7

HIGH

2020-10-28

Bug fixes

• Fixed an issue where the SETRANGE command exited abnormally.

• Fixed a data exception caused by an empty key string.

1.2.6

HIGH

2020-09-28

Bug fixes

• Fixed an issue where connection management logs grew too quickly.

1.2.5

HIGH

2020-09-27

Bug fixes

• Fixed an issue where operational logs grew too quickly.

1.2.4

HIGH

2020-09-27

Feature optimizations

• Improved the stability of primary-secondary synchronization.

• Optimized the performance of scanning members of complex data structures.

Bug fixes

• Fixed an issue where the ZADD and ZINCR commands were inconsistent with the native commands (semantic error when operating on multiple duplicate members).

• Fixed an error where the MGET command returned nil for an empty string.

1.2.3

LOW

2020-09-27

Initial release

Tair (Enterprise Edition) Disk-based is independently developed based on the TairDB KV storage engine and uses enterprise SSD (ESSD) for storage. It provides a large-capacity, low-cost, and highly persistent database service with the following advantages:

• All data is stored on disks, providing natural persistence at a cost as low as 15% of Redis Open-Source Edition.

• It uses a multi-I/O, multi-worker model to improve single-machine throughput and supports primary-secondary replication with binary logs.

• It is compatible with most native Redis commands.

• It resolves the issue of reserving memory due to the fork problem inherent in native Redis.

Minor version

Update level

Release date

Type

Description

1.2.7

HIGH

2024-07-10

New features

• The pubsub_clients metric is added to the INFO CLIENTS command.

• Traffic can be split into data traffic and replication traffic.

Feature optimizations

• The efficiency of active expiration is optimized.

• The performance of CLUSTER SLOTS and CLUSTER NODES is optimized.

• Optimized the statistics of used memory to avoid false memory watermark warnings.

• Statistics about hotkeys are regularly recorded in audit logs.

Bug fixes

• The CVE-2022-24834 security vulnerability is fixed.

• Fixed an issue where TairHash could occasionally crash.

• Fixed an issue where historical hot key information was incomplete by periodically outputting hot key information.

• Improved the stability of cluster scaling.

• Improved the stability of semi-synchronous replication.

1.2.6.4

LOW

2023-10-13

Bug fixes

• Improved the stability of semi-synchronous replication.

1.2.6.3

MEDIUM

2023-08-03

Bug fixes

• Fixed an issue where the eviction feature was abnormal in special scenarios.

• Improved stability.

1.2.6.2

LOW

2022-12-09

Bug fixes

• Improved stability.

1.2.6.1

LOW

2022-11-14

Bug fixes

• Improved stability.

1.2.6

LOW

2022-11-02

New features

• Added support for TairHash.

Feature optimizations

• Optimized request execution performance.

• Improved stability.

1.2.5.1

LOW

2022-09-21

Bug fixes

• Fixed a bug related to TairCPC commands in version 1.2.5, which could cause instance failures.

  Note

  Upgrade instances of version 1.2.5 to this version.

• Improved stability.

1.2.5

MEDIUM

2022-08-31

Feature optimizations

• Optimized the main hash table extension strategy.

• Optimized the performance of the FLUSHALL and FLUSHDB commands.

• Optimized Rehash performance.

• Improved stability.

1.2.4.1

LOW

2022-07-28

Bug fixes

• Fixed a bug in version 1.2.4 where the GETBIT command could cause instance failures.

  Note

  Upgrade instances of version 1.2.4 to this version.

1.2.4

LOW

2022-07-18

Feature optimizations

• Compatible with RENAME series commands and the MOVE command. The time complexity is O(n), which is different from the original version. This can be controlled by the pena_rename_move_compatible_enabled parameter.

• Improved the stability of cluster instance scaling.

• Optimized the performance of the SETBIT, SETRANGE, APPEND, and BITFIELD commands to resolve slow execution on large keys.

1.2.3.4

LOW

2022-07-14

Bug fixes

• Removed the Errorstats - Selected field from the INFO command return.

• Optimized the latency statistics histogram.

• Improved the stability of semi-synchronous replication.

• Fixed an issue where commands such as ZUNIONSTORE and ZINTERSTORE reported errors in cluster proxy mode.

• Fixed an issue where the SORT command could cause incorrect persistent memory usage statistics in special scenarios.

Feature optimizations

• The redis_version return value for the INFO command is changed to 6.0.9.

1.2.3.3

LOW

2022-05-12

New features

• Added support for TairCpc.

1.2.3.2

HIGH

2022-04-24

Bug fixes

• Fixed an issue where the order of key eviction was abnormal in MAXMEMORY_VOLATILE_TTL mode.

• Upgraded and optimized the process of adding and deleting nodes in a cluster architecture.

• Improved the stability of semi-synchronous replication.

1.2.3.1

LOW

2022-03-31

Feature optimizations

• Fixed a persistence issue with the CAS command.

• Fixed an issue where the client IP address was not displayed in the slow query logs and audit logs of cluster instances.

• Improved stability.

1.2.3

MEDIUM

2022-03-03

New features

• Added support for TairString.

• Added support for the Stream data structure.

• The Set structure now supports the intset structure, reducing memory overhead.

• Added support for data eviction (Eviction policy).

• Optimized performance and improved stability.

Bug fixes

• Fixed an issue where the real-time key analysis feature missed statistics for the Spop command.

1.2.2.4

LOW

2022-01-21

Feature optimizations

• Optimized storage space for large value scenarios to reduce persistent memory usage.

1.2.2.3

HIGH

2021-12-30

Feature optimizations

• Optimized tail latency when writing data.

• Improved stability.

1.2.2.2

HIGH

2021-12-14

Bug fixes

• Fixed an issue where usage statistics for structures such as List and Hash were incorrect in special scenarios.

1.2.2.1

LOW

2021-10-21

Feature optimizations

• Improved stability.

1.2.2

HIGH

2021-10-20

Feature updates

• Optimized the speed control of automatic scaling for cluster instances.

  Note

  During scaling, the cleanup speed of old data is adaptively controlled based on the instance load. For example, more CPU is used for cleanup during low load to speed up the process. This feature causes the CPU usage to be at a high level during scaling, but it does not affect normal access.

• Optimized the performance of commands such as CLUSTER NODES in large-scale clusters.

Bug fixes

• Fixed an issue where abnormal slow query logs (SLOWLOG) were generated during cluster scaling.

• Improved stability.

1.2.0

LOW

2021-09-21

Feature updates

• Optimized storage space for List, Hash, Set, and Zset structures to reduce memory and persistent memory usage.

1.1.8

LOW

2021-08-17

Feature updates

• Resolved an issue where memory exhaustion prevented full utilization of persistent memory in some scenarios.

• Improved stability.

1.1.7

LOW

2021-08-02

New features

• The semi-synchronous feature is supported, but it is disabled by default.

  Note

  Semi-synchronous means that after an update initiated by a client is executed on the primary node, the log is synchronously transmitted to the secondary node. The primary node returns to the client only after the secondary node confirms receipt. When the secondary instance is unavailable or a network exception occurs between the primary and secondary instances, semi-synchronous replication degrades to asynchronous.

• In the return information of the INFO command, the redis_version return value is changed to 4.9.9, and pena_version is added as the database engine version (minor version).

1.1.6.1

MEDIUM

2021-06-10

Feature updates

• Improved stability.

1.1.6

MEDIUM

2021-05-08

New features

• Supports seamless scaling, which enables the seamless migration of slots.

Feature optimizations

• Hash and Zset structures now support ziplist encoding, reducing memory overhead.

• Enhanced the liveness detection capability of the high availability (HA) system.

• Optimized the migration capability of large clusters (that is, clusters with many data shards).

• Improved the stability of scaling in a cluster architecture.

1.1.5

MEDIUM

2021-01-15

New features

• Supports cluster architecture to overcome the single-threaded bottleneck of Redis and meet requirements for high capacity and performance.

• Uses the minor version of the instance as the value of the redis_version parameter returned by the INFO command.

Feature optimizations

• Improved service stability.

1.1.4

MEDIUM

2020-10-28

New features

• When a standard account runs the INFO command, the output now includes persistent memory usage information.

Feature optimizations

• Optimized the memory space that the EMBSTR encoding method uses to reduce memory overhead.

• Improved service stability.

1.1.3

LOW

2020-09-16

Initial release

Tair (Enterprise Edition) Persistent Memory-optimized is based on persistent memory technology. It provides a large-capacity, Redis-compatible in-memory database with the following advantages:

• Costs about 30% less than Redis Open-Source Edition for the same capacity.

• Resolves issues such as latency jitter and slow data loading caused by fork calls during an AOF rewrite on large instances. This eliminates the need to choose between performance and persistence.

• Provides strong command-level persistence. A success response is returned only after each write operation is persisted.

• Compatible with most data structures and interfaces of native Redis. It also supports persistence for data structures, except for Streams.