Release time : 2020-01-07
Version compatible
Upgrade must-see
-
load_collection()
Method supports pre-loading of designated partitions #4307
Major improvements
-
Optimized the process of system indexing and query. #4454
Problem fix
-
Calling
load_collection()
andsearch()
methods in multiple threads will cause Milvus to crash. #4378 -
Milvus only searches the default partition when the search parameter
partition_tags
contains_default
words. #4484
See CHANGELOG for more information about fixed issues!
The Milvus vector search engine can help users easily deal with massive unstructured data (picture/video/voice/text) retrieval. A single node Milvus can complete a billion-level vector search in seconds (please refer to: online tutorial ), and the distributed architecture can also meet the horizontal expansion needs of users.
The application scenarios of the Milvus volume search engine include: Internet entertainment (image search/video search), new retail (search for goods by image), smart finance (user authentication), and smart logistics (vehicle identification).