Scalar Indexes in LanceDB

Scalar indexes organize data by scalar attributes (e.g., numbers, categories) and enable fast filtering of vector data. They accelerate retrieval of scalar data associated with vectors, thus enhancing query performance.

LanceDB supports three types of scalar indexes:

BTREE: Stores column data in sorted order for binary search. Best for columns with many unique values.
BITMAP: Uses bitmaps to track value presence. Ideal for columns with few unique values (e.g., categories, tags).
LABEL_LIST: Special index for List<T> columns supporting array_contains_all and array_contains_any queries.

Choosing the Right Index Type

Data Type	Filter	Index Type
Numeric, String, Temporal	`<`, `=`, `>`, `in`, `between`, `is null`	`BTREE`
Boolean, numbers or strings with fewer than 1,000 unique values	`<`, `=`, `>`, `in`, `between`, `is null`	`BITMAP`
List of low cardinality of numbers or strings	`array_has_any`, `array_has_all`	`LABEL_LIST`

Scalar Index Operations

1. Build the Index

You can create multiple scalar indexes within a table. By default, the index will be BTREE, but you can always configure another type like BITMAP

import lancedb

# Connect to LanceDB
db = lancedb.connect("data")
table = db.open_table("my_vectors")

# Create BTREE index (default)
table.create_scalar_index("book_id")

# Create BITMAP index
table.create_scalar_index("publisher", index_type="bitmap")

import * as lancedb from "@lancedb/lancedb"

const db = await lancedb.connect("data");
const tbl = await db.openTable("my_vectors");

await tbl.createIndex("book_id");
await tbl.createIndex("publisher", { config: lancedb.Index.bitmap() })

💡 LanceDB Cloud and Enterprise

If you are using Cloud or Enterprise, the create_scalar_index API returns immediately, but the building of the scalar index is asynchronous. To wait until all data is fully indexed, you can specify the wait_timeout parameter on create_scalar_index() or call wait_for_index() on the table.

2. Check Index Status

You can use the UI Dashboard in Cloud or just call the API:

index_name = "label_idx"
table.wait_for_index([index_name])

const indexName = "label_idx"
await table.waitForIndex([indexName], 60)

3. Update the Index

Updating the table data (adding, deleting, or modifying records) requires that you also update the scalar index. This can be done by calling optimize, which will trigger an update to the existing scalar index.

# Add new data
table.add([{"vector": [7, 8], "book_id": 4}])

# Optimize to update indexes
table.optimize()

await tbl.add([{ vector: [7, 8], book_id: 4 }]);
await tbl.optimize();

💡 LanceDB Cloud

New data added after creating the scalar index will still appear in search results if optimize is not used, but with increased latency due to a flat search on the unindexed portion. LanceDB Cloud automates the optimize process, minimizing the impact on search speed.

4. Run Indexed Search

The following scan will be faster if the column book_id has a scalar index:

import lancedb

# Connect to LanceDB
db = lancedb.connect("data")
table = db.open_table("books")

# Query with scalar index
result = table.search().where("book_id = 2").limit(10).to_pandas()

const db = await lancedb.connect("data");
const tbl = await db.openTable("books");

await tbl
  .query()
  .where("book_id = 2")
  .limit(10)
  .toArray();

Scalar indexes can also speed up scans containing a vector search or full text search, and a prefilter:

import lancedb

# Connect to LanceDB
db = lancedb.connect("data")
table = db.open_table("book_with_embeddings")

# Vector search with scalar prefilter
result = table.search([1.2] * 1536) \
    .where("book_id != 3") \
    .limit(10) \
    .to_pandas()

const db = await lancedb.connect("data/lance");
const tbl = await db.openTable("book_with_embeddings");

await tbl.search(Array(1536).fill(1.2))
  .where("book_id != 3")  // prefilter is default behavior.
  .limit(10)
  .toArray();

Advanced: Index UUID Columns

LanceDB supports scalar indexes on UUID columns (stored as FixedSizeBinary(16)), enabling efficient lookups and filtering on UUID-based primary keys.

💡 FixedSizeBinary Requirements

To use FixedSizeBinary, ensure you have:

Python SDK version 0.22.0-beta.4 or later
TypeScript SDK version 0.19.0-beta.4 or later

1. Define UUID Type

First, create a custom Arrow extension type for UUIDs:

import pyarrow as pa

class UuidType(pa.ExtensionType):
    def __init__(self):
        super().__init__(pa.binary(16), "my.uuid")

    def __arrow_ext_serialize__(self):
        return b'uuid-metadata'

    @classmethod
    def __arrow_ext_deserialize__(cls, storage_type, serialized):
        return UuidType()

pa.register_extension_type(UuidType())

import { Schema, Field, FixedSizeBinary } from "apache-arrow"

// UUID is stored as FixedSizeBinary(16)
const schema = new Schema([
    new Field("id", new FixedSizeBinary(16), true),
    new Field("name", new Utf8(), false),
]);

2. Generate UUID Data

Create helper functions to generate UUIDs and sample data:

import uuid
import random
import string

def generate_random_string(length=10):
    return ''.join(random.choices(string.ascii_letters + string.digits, k=length))

def generate_uuids(num_items):
    return [uuid.uuid4().bytes for _ in range(num_items)]

# Generate sample data
n = 100
uuids = generate_uuids(n)
names = [generate_random_string() for _ in range(n)]

import { v4 as uuidv4 } from "uuid"

function uuidToBuffer(uuid: string): Buffer {
    return Buffer.from(uuid.replace(/-/g, ''), 'hex');
}

function generateUuids(numItems: number): Buffer[] {
    return Array.from({ length: numItems }, () => uuidToBuffer(uuidv4()));
}

function generateRandomString(length: number = 10): string {
    const chars = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789';
    return Array.from(
        { length },
        () => chars.charAt(Math.floor(Math.random() * chars.length))
    ).join('');
}

// Generate sample data
const n = 100;
const uuids = generateUuids(n);
const names = Array.from({ length: n }, () => generateRandomString());

3. Create Table with UUID Column

Create a table with the UUID column and sample data:

import lancedb

# Connect to LanceDB
db = lancedb.connect(
    uri="db://your-project-slug",
    api_key="your-api-key",
    region="us-east-1"
)

# Create arrays
uuid_array = pa.array(uuids, pa.binary(16))
name_array = pa.array(names, pa.string())
extension_array = pa.ExtensionArray.from_storage(UuidType(), uuid_array)

# Create schema
schema = pa.schema([
    pa.field('id', UuidType()),
    pa.field('name', pa.string())
])

# Create table
data_table = pa.Table.from_arrays([extension_array, name_array], schema=schema)
table_name = "index-on-uuid-test"
table = db.create_table(table_name, data=data_table, mode="overwrite")

import * as lancedb from "@lancedb/lancedb"

// Connect to LanceDB
const db = await lancedb.connect({
    uri: "db://your-project-slug",
    apiKey: "your-api-key",
    region: "us-east-1"
});

// Create data array
const data = makeArrowTable(
    uuids.map((id, index) => ({
        id,
        name: names[index],
    })),
    { schema }
);

// Create table
const tableName = "index-on-uuid-test-ts";
const table = await db.createTable(tableName, data, { mode: "overwrite" });

4. Create and Wait for Index

Create a scalar index on the UUID column and wait for it to complete:

# Create index
table.create_scalar_index("id")
index_name = "id_idx"
wait_for_index(table, index_name)

// Create scalar index
await table.createIndex("id", {
    config: Index.bitmap(),
});

// Wait for index
const scalarIndexName = "id_idx";
await waitForIndex(table, scalarIndexName);

5. Perform Operations with UUID Index

Example of upserting new records with UUIDs:

# Upsert example
new_users = [
    {"id": uuid.uuid4().bytes, "name": "Bobby"},
    {"id": uuid.uuid4().bytes, "name": "Charlie"},
]

table.merge_insert("id") \
    .when_matched_update_all() \
    .when_not_matched_insert_all() \
    .execute(new_users)

// Upsert example
const newUsers = [
    { id: uuidToBuffer(uuidv4()), name: "Bobby" },
    { id: uuidToBuffer(uuidv4()), name: "Charlie" },
];

await table.mergeInsert("id")
    .whenMatchedUpdateAll()
    .whenNotMatchedInsertAll()
    .execute(newUsers);