Indexes

On this page Carat arrow pointing down

Indexes improve your database's performance by helping SQL locate data without having to look through every row of a table.

How do indexes work?

When you create an index, CockroachDB "indexes" the columns you specify, which creates a copy of the columns and then sorts their values (without sorting the values in the table itself).

After a column is indexed, SQL can easily filter its values using the index instead of scanning each row one-by-one. On large tables, this greatly reduces the number of rows SQL has to use, executing queries exponentially faster.

For example, if you index an INT column and then filter it WHERE <indexed column> = 10, SQL can use the index to find values starting at 10 but less than 11. In contrast, without an index, SQL would have to evaluate every row in the table for values equaling 10. This is also known as a "full table scan", and it can be very bad for query performance.

Creation

Each table automatically has an index created called primary, which indexes either its primary key or—if there is no primary key—a unique value for each row known as rowid. We recommend always defining a primary key because the index it creates provides much better performance than letting CockroachDB use rowid.

The primary index helps filter a table's primary key but doesn't help SQL find values in any other columns. However, you can use secondary indexes to improve the performance of queries using columns not in a table's primary key. You can create them:

  • At the same time as the table with the INDEX clause of CREATE TABLE. In addition to explicitly defined indexes, CockroachDB automatically creates secondary indexes for columns with the UNIQUE constraint.
  • For existing tables with CREATE INDEX.
  • By applying the UNIQUE constraint to columns with ALTER TABLE, which automatically creates an index of the constrained columns.

To create the most useful secondary indexes, you should also check out our best practices.

Selection

Because each query can use only a single index, CockroachDB selects the index it calculates will scan the fewest rows (i.e., the fastest). For more detail, check out our blog post Index Selection in CockroachDB, which will show you how to use the EXPLAIN statement for your query to see which index is being used.

To override CockroachDB's index selection, you can also force queries to use a specific index (also known as "index hinting"). Index hinting is supported for SELECT, DELETE, and UPDATE statements.

Storage

CockroachDB stores indexes directly in your key-value store. You can find more information in our blog post Mapping Table Data to Key-Value Storage.

Locking

Tables are not locked during index creation thanks to CockroachDB's schema change procedure.

Performance

Indexes create a trade-off: they greatly improve the speed of queries, but may slightly slow down writes to an affected column (because new values have to be written for both the table and the index).

To maximize your indexes' performance, we recommend following a few best practices.

Hash-sharded indexes

Warning:

This is an experimental feature. The interface and output are subject to change.

New in v20.1: CockroachDB automatically splits ranges of data in the key-value store based on the size of the range, and on the load streaming to the range. To split a range based on load, the system looks for a point in the range that evenly divides incoming traffic. If the range is indexed on a column of data that is sequential in nature (e.g., an ordered sequence, or a series of increasing, non-repeating TIMESTAMPs), then all incoming writes to the range will be the last (or first) item in the index and appended to the end of the range. As a result, the system cannot find a point in the range that evenly divides the traffic, and the range cannot benefit from load-based splitting, creating a hotspot on the single range.

If you are working with a table that must be indexed on sequential keys, you should use hash-sharded indexes. Hash-sharded indexes distribute sequential traffic uniformly across ranges, eliminating single-range hotspots and improving write performance on sequentially-keyed indexes at a small cost to read performance. For details about the mechanics and performance improvements of hash-sharded indexes in CockroachDB, see our Hash Sharded Indexes Unlock Linear Scaling for Sequential Workloads blog post.

To create a hash-sharded index, set the experimental_enable_hash_sharded_indexes session variable to on. Then, add the optional USING HASH WITH BUCKET_COUNT = n_buckets clause to a CREATE INDEX statement, to an INDEX definition in a CREATE TABLE statement, or to an ALTER PRIMARY KEY statement. When this clause is used, CockroachDB creates n_buckets computed columns, shards the index into n_buckets shards, and then stores each index shard in the underlying key-value store with one of the computed column's hash as its prefix.

To change the bucket size of an existing hash-sharded primary key index, use an ALTER PRIMARY KEY statement with a USING HASH WITH BUCKET_COUNT = n_buckets clause that specifies the new bucket size and the existing primary key columns.

Note:

Hash-sharded indexes cannot be interleaved.

Best practices

We recommend creating indexes for all of your common queries. To design the most useful indexes, look at each query's WHERE and SELECT clauses, and create indexes that:

Tip:

For more information about how to tune CockroachDB's performance, see SQL Performance Best Practices and the Performance Tuning tutorial.

Indexing columns

When designing indexes, it's important to consider which columns you index and the order in which you list them. Here are a few guidelines to help you make the best choices:

  • Queries can benefit from an index even if they only filter a prefix of its columns. For example, if you create an index of columns (A, B, C), queries filtering (A) or (A, B) can still use the index. However, queries that do not filter (A) will not benefit from the index.

    This feature also lets you avoid using single-column indexes. Instead, use the column as the first column in a multiple-column index, which is useful to more queries.
  • Columns filtered in the WHERE clause with the equality operators (= or IN) should come first in the index, before those referenced with inequality operators (<, >).
  • Indexes of the same columns in different orders can produce different results for each query. For more information, see our blog post on index selection—specifically the section "Restricting the search space."
  • Avoid indexing on sequential values. Writes to indexes with sequential keys can result in range hotspots that negatively affect performance. Instead, use randomly generated unique IDs, or multi-column keys.
  • Avoid creating secondary indexes that you do not need, as they can slow down write performance and take up node memory. For example, if you want to change a primary key, and you do not plan to filter queries on the old primary key column(s), do not use ALTER PRIMARY KEY, which creates a secondary index from the old primary key. Instead, use DROP CONSTRAINT ... PRIMARY KEY/ADD CONSTRAINT ... PRIMARY KEY, which does not create a secondary index.

Storing columns

The STORING clause specifies columns which are not part of the index key but should be stored in the index. This optimizes queries which retrieve those columns without filtering on them, because it prevents the need to read the primary index.

Example

Say we have a table with three columns, two of which are indexed:

icon/buttons/copy
> CREATE TABLE tbl (col1 INT, col2 INT, col3 INT, INDEX (col1, col2));

If we filter on the indexed columns but retrieve the unindexed column, this requires reading col3 from the primary index via an "index join."

icon/buttons/copy
> EXPLAIN SELECT col3 FROM tbl WHERE col1 = 10 AND col2 > 1;
       tree       |    field    |      description
------------------+-------------+------------------------
                  | distributed | false
                  | vectorized  | false
  render          |             |
   └── index-join |             |
        │         | table       | tbl@primary
        │         | key columns | rowid
        └── scan  |             |
                  | table       | tbl@tbl_col1_col2_idx
                  | spans       | /10/2-/11
(9 rows)

However, if we store col3 in the index, the index join is no longer necessary. This means our query only needs to read from the secondary index, so it will be more efficient.

icon/buttons/copy
> CREATE TABLE tbl (col1 INT, col2 INT, col3 INT, INDEX (col1, col2) STORING (col3));
icon/buttons/copy
> EXPLAIN SELECT col3 FROM tbl WHERE col1 = 10 AND col2 > 1;
    tree    |    field    |      description
------------+-------------+------------------------
            | distributed | false
            | vectorized  | false
  render    |             |
   └── scan |             |
            | table       | tbl@tbl_col1_col2_idx
            | spans       | /10/2-/11
(6 rows)

See also


Yes No
On this page

Yes No