Indeks Peliput dan Pemindaian Hanya Indeks

Welcome to our lesson on Covering Indexes and Index-Only Scans! These advanced indexing techniques can dramatically boost your PostgreSQL query performance.

A regular index helps PostgreSQL quickly find rows based on certain column values. Think of it like a book's index: it tells you which page to go to for a topic.

When you use a standard index, PostgreSQL first finds the row's location (its tuple ID or CTID) in the index. Then, it has to go to the actual table to fetch the rest of the data for that row.

This extra step, called a table lookup or heap fetch, can be slow, especially if many rows are involved or if the table data isn't in memory.

A covering index is an index that contains all the columns needed to fulfill a query, not just the columns in the WHERE clause. This means PostgreSQL can answer the query by reading only the index, without ever touching the main table data.

It 'covers' the query's data requirements entirely.

To create a covering index, you include all the columns your query might select or filter on. PostgreSQL supports including non-key columns using the INCLUDE clause, which stores them without making them part of the primary sort key.

Let's create a table and then a covering index.

When a query can be fully satisfied by a covering index, PostgreSQL performs an Index-Only Scan. This is a highly efficient operation because the database engine doesn't need to read any data blocks from the main table (the 'heap').

It's like finding all the information you need directly in the book's index, without ever turning to the main chapters.

PostgreSQL uses MVCC (Multi-Version Concurrency Control). For an Index-Only Scan to work, PostgreSQL must ensure that the versions of the rows it finds in the index are visible to the current transaction.

It does this by checking the table's visibility map, a special data structure that tracks which table pages contain only 'all-visible' tuples. If a page is marked all-visible, no heap fetch is needed to check row visibility.

You can verify if PostgreSQL is using an Index-Only Scan by checking the query plan with EXPLAIN ANALYZE. Look for Index Only Scan in the output.

Let's try a query that our covering index should handle.

Index-Only Scans offer several key advantages:

• Reduced Disk I/O: Fewer disk reads are needed because the main table is skipped.
• Faster Query Execution: Less I/O directly translates to quicker query responses.
• Better Cache Utilization: Indexes are often smaller and more frequently accessed, leading to better use of memory caches.
• Concurrency: Less contention on table data blocks.

While powerful, covering indexes aren't a silver bullet:

• Larger Indexes: Including more columns makes the index physically larger, consuming more disk space.
• Slower Writes: Updates, inserts, and deletes to the main table also require updating the index, which can be slower for larger indexes.
• Maintenance: Larger indexes might take longer to VACUUM.

Use them strategically for read-heavy queries that frequently access a specific subset of columns.

Consider a `users` table where you often query user names and emails based on their registration date. An index on `registration_date` including `name` and `email` would be a perfect candidate for a covering index.

You've learned that covering indexes include all columns needed by a query, enabling highly efficient Index-Only Scans. These scans bypass the main table, significantly reducing disk I/O and speeding up read-heavy queries.

Remember to use EXPLAIN ANALYZE to confirm their usage and consider the trade-offs of increased index size and potential write overhead. Happy optimizing!