Manage Indexes

All ANN searches on Zilliz Cloud rely on indexes for extremely high performance. Before any ANN searches, you have to create indexes on your collection and load the indexes into CUs.

Create an index

The following example creates an index on the title_vector field in the medium_article_2020 collection created in this guide.

from pymilvus import Collection

index_params = {
    "index_type": "AUTOINDEX",
    "metric_type": "L2",
    "params": {}
}

collection = Collection("medium_articles_2020")
collection.create_index(
  field_name="title_vector", 
  index_params=index_params
)

AUTOINDEX is a proprietary index type available on Zilliz Cloud for index auto-optimization.

Currently, Zilliz Cloud does not allow indexes on scalar fields.

Load an index

Load the index of the collection for ANN searches.

from pymilvus import Collection

collection = Collection("medium_articles_2020")      
collection.load()

Release an index

For some collections that are not frequently used, you can release their indexes to save CUs. You can load them again later when needed. Note that this operation only releases indexes from CUs, but does not delete data or its index from the collection.

from pymilvus import Collection

collection = Collection("medium_articles_2020")     
collection.release()

Since reloading a collection is time-consuming, only release the collections that will stay idle for a long time, e.g. more than 10 hours.

Drop an index

For those indexes no longer in need, drop them as follows:

from pymilvus import Collection

collection = Collection("medium_articles_2020")     
collection.drop_index()

Note that, if you want to change index parameters, such as changing metric_type from L2 to IP, you need to drop the current index and create a new one.

Next steps

On this page