惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Hacker News: Ask HN
Hacker News: Ask HN
C
Cisco Blogs
The Hacker News
The Hacker News
T
Tor Project blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
The GitHub Blog
The GitHub Blog
A
Arctic Wolf
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
The Register - Security
The Register - Security
云风的 BLOG
云风的 BLOG
Simon Willison's Weblog
Simon Willison's Weblog
P
Palo Alto Networks Blog
Vercel News
Vercel News
C
CERT Recently Published Vulnerability Notes
I
InfoQ
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
M
MIT News - Artificial intelligence
I
Intezer
aimingoo的专栏
aimingoo的专栏
U
Unit 42
C
Cyber Attacks, Cyber Crime and Cyber Security
L
LINUX DO - 热门话题
Microsoft Security Blog
Microsoft Security Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
Cyberwarzone
Cyberwarzone
P
Proofpoint News Feed
P
Proofpoint News Feed
B
Blog
T
Threat Research - Cisco Blogs
博客园 - 叶小钗
Recorded Future
Recorded Future
Last Week in AI
Last Week in AI
N
News and Events Feed by Topic
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
Know Your Adversary
Know Your Adversary
Engineering at Meta
Engineering at Meta
G
Google Developers Blog
PCI Perspectives
PCI Perspectives
Google DeepMind News
Google DeepMind News
WordPress大学
WordPress大学
Application and Cybersecurity Blog
Application and Cybersecurity Blog
MyScale Blog
MyScale Blog
Security Archives - TechRepublic
Security Archives - TechRepublic
Schneier on Security
Schneier on Security
N
News | PayPal Newsroom
C
Cybersecurity and Infrastructure Security Agency CISA
H
Help Net Security
博客园 - 聂微东
H
Hackread – Cybersecurity News, Data Breaches, AI and More
G
GRAHAM CLULEY

Blog — PlanetScale

Keeping a Postgres queue healthy — PlanetScale Patterns for Postgres Traffic Control — PlanetScale Graceful degradation in Postgres — PlanetScale High memory usage in Postgres is good, actually — PlanetScale Stripe Projects partnership: Provision PlanetScale Postgres and MySQL databases from the Stripe CLI — PlanetScale Enhanced tagging in Postgres Query Insights — PlanetScale Behind the scenes: How Database Traffic Control works — PlanetScale Introducing Database Traffic Control — PlanetScale Scaling Postgres connections with PgBouncer — PlanetScale Drizzle joins PlanetScale — PlanetScale Video Conferencing with Postgres — PlanetScale Faster PlanetScale Postgres connections with Cloudflare Hyperdrive — PlanetScale Introducing the PlanetScale MCP server — PlanetScale Database Transactions — PlanetScale Automating our changelog with Cursor commands — PlanetScale Postgres 18 is now available — PlanetScale Using MotherDuck with PlanetScale — PlanetScale $50 PlanetScale Metal is GA for Postgres — PlanetScale AI-Powered Postgres index suggestions — PlanetScale $5 PlanetScale is live — PlanetScale Announcing Vitess 23 — PlanetScale $50 PlanetScale Metal — PlanetScale Report on our investigation of the 2025-10-20 incident in AWS us-east-1 — PlanetScale $5 PlanetScale — PlanetScale Benchmarking Postgres 17 vs 18 — PlanetScale Larger than RAM Vector Indexes for Relational Databases — PlanetScale Partnering with Cloudflare to bring you the fastest globally distributed applications — PlanetScale Processes and Threads — PlanetScale PlanetScale for Postgres is now GA — PlanetScale Postgres High Availability with CDC — PlanetScale Announcing Neki — PlanetScale Caching — PlanetScale The principles of extreme fault tolerance — PlanetScale Announcing PlanetScale for Postgres — PlanetScale Benchmarking Postgres — PlanetScale Announcing Vitess 22 — PlanetScale The Real Failure Rate of EBS — PlanetScale IO devices and latency — PlanetScale Announcing PlanetScale Metal — PlanetScale PlanetScale Metal: There’s no replacement for displacement — PlanetScale Upgrading Query Insights to Metal — PlanetScale Automating cherry-picks between OSS and private forks — PlanetScale Database Sharding — PlanetScale Anatomy of a Throttler, part 3 — PlanetScale Introducing sharding on PlanetScale with workflows — PlanetScale Announcing Vitess 21 — PlanetScale Announcing the PlanetScale vectors public beta — PlanetScale Anatomy of a Throttler, part 2 — PlanetScale Instant deploy requests — PlanetScale Anatomy of a Throttler, part 1 — PlanetScale Increase IOPS and throughput with sharding — PlanetScale Tracking index usage with Insights — PlanetScale Faster backups with sharding — PlanetScale Building data pipelines with Vitess — PlanetScale The State of Online Schema Migrations in MySQL — PlanetScale Optimizing aggregation in the Vitess query planner — PlanetScale Dealing with large tables — PlanetScale Announcing Vitess 20 — PlanetScale Self-managed Vitess vs Managed Vitess with PlanetScale — PlanetScale Achieving data consistency with the consistent lookup Vindex — PlanetScale The MySQL adaptive hash index — PlanetScale Introducing global replica credentials — PlanetScale Profiling memory usage in MySQL — PlanetScale Summer 2023: Fuzzing Vitess at PlanetScale — PlanetScale How PlanetScale makes schema changes — PlanetScale Identifying and profiling problematic MySQL queries — PlanetScale The Problem with Using a UUID Primary Key in MySQL — PlanetScale Announcing Vitess 19 — PlanetScale PlanetScale forever — PlanetScale Introducing schema recommendations — PlanetScale Amazon Aurora Pricing: The many surprising costs of running an Aurora database — PlanetScale Three common MySQL database design mistakes — PlanetScale OAuth applications are now available to everyone — PlanetScale Deprecating the Scaler plan — PlanetScale PlanetScale branching vs. Amazon Aurora blue/green deployments — PlanetScale Databases at scale — PlanetScale Considerations for building a database disaster recovery plan — PlanetScale Working with Geospatial Features in MySQL — PlanetScale PlanetScale vs Amazon Aurora replication — PlanetScale Introducing the Vantage and PlanetScale integration — PlanetScale MySQL isolation levels and how they work — PlanetScale Introducing the schemadiff command line tool — PlanetScale $ pscale ping — PlanetScale Announcing foreign key constraints support — PlanetScale The challenges of supporting foreign key constraints — PlanetScale What is HTAP? — PlanetScale Introducing Insights Anomalies — PlanetScale Webhook security: a hands-on guide — PlanetScale Announcing the Fivetran integration — PlanetScale Introducing webhooks — PlanetScale What is MySQL replication and when should you use it? — PlanetScale Sync user data between Clerk and a PlanetScale MySQL database — PlanetScale Introducing database reports — PlanetScale Distributed caching systems and MySQL — PlanetScale What is MySQL partitioning? — PlanetScale MySQL High Availability: Connection handling and concurrency — PlanetScale Personalizing your onboarding with Markdoc — PlanetScale PlanetScale vs. Amazon Aurora — PlanetScale PlanetScale vs. Amazon RDS — PlanetScale PlanetScale is bringing vector search and storage to MySQL — PlanetScale
Using redundant conditions to unlock indexes in MySQL — PlanetScale
Aaron Francis · 2023-06-07 · via Blog — PlanetScale

Aaron Francis |

When working with MySQL (or any database!), it's essential to understand how indexes work and how they can be used to improve the efficiency of queries. An index is a separate data structure that maintains a copy of part of your data, structured to allow quick data retrieval. Usually, this structure is a B+ Tree. We have an entire post on how indexes work if you want to go into greater detail.

Obfuscated indexes

Creating indexes is only part of the battle. You must also know how to write your queries so that you allow MySQL to use your indexes. One common mistake people make when writing queries is that they obfuscate their indexes. Obfuscating an index simply means that you're hiding the indexed value from MySQL.

Let's say you have a todos table with a created_at column that records a timestamp of when the record was created.

CREATE TABLE `todos` (
  `id` int NOT NULL AUTO_INCREMENT,
  `title` varchar(255) NOT NULL,
  `created_at` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
  PRIMARY KEY (`id`),
  KEY `created_at` (`created_at`)
)

In this table, we've added an index to the created_at column to quickly filter by that timestamp. When we query against the created_at column to find records created in the last 24 hours, we see that MySQL is using the index as we'd expect:

EXPLAIN SELECT * FROM todos WHERE created_at > NOW() - INTERVAL 24 HOUR;

-- | id | type  | possible_keys | key        | key_len | ref | rows | filtered | Extra                 |
-- |----|-------|---------------|------------|---------|-----|------|----------|-----------------------|
-- |  1 | range | created_at    | created_at | 4       |     |    1 |   100.00 | Using index condition |

However, if we wrap this column in a function, we're obfuscating the column from MySQL, and it can no longer use the index.

EXPLAIN SELECT * FROM todos WHERE YEAR(created_at) = 2023;

-- | id | type | possible_keys | key | key_len | ref | rows  | filtered | Extra       |
-- |----|------|---------------|-----|---------|-----|-------|----------|-------------|
-- |  1 | ALL  |               |     |         |     | 39746 |   100.00 | Using where |

By wrapping the created_at column in a YEAR function, we're asking MySQL to do an index lookup on YEAR(created_at), which is not an index MySQL maintains. It is only maintaining the created_at index.

In some cases, there are ways around index obfuscation. In this example, we could use a range scan instead of the YEAR function to obtain the same result.

EXPLAIN SELECT * FROM todos WHERE created_at BETWEEN '2023-01-01 00:00:00' AND '2023-12-31 23:59:59';

-- | id | type  | possible_keys | key        | key_len | ref | rows | filtered | Extra                 |
-- |----|-------|---------------|------------|---------|-----|------|----------|-----------------------|
-- |  1 | range | created_at    | created_at | 4       |     |    1 |   100.00 | Using index condition |

By unwrapping the created_at column and changing the comparison to a range scan, we've unlocked the index and allowed MySQL to use it effectively.

Unfortunately, it's not always possible to de-obfuscate your indexes. In some scenarios, you simply cannot avoid wrapping the column in a function. In these cases, you might see if there is a redundant condition that could potentially unlock an existing index.

Redundant conditions in MySQL

A redundant condition is a condition that seems superfluous, extra, or not needed. It is a condition that can be added and removed without changing the results that MySQL returns.

Let's take a look at a contrived example to illustrate the point. In this example, we're selecting the todos with an id of less than five.

SELECT
  *
FROM
  todos
WHERE
  id < 5

In this case, a redundant condition might be id < 10.

SELECT
  *
FROM
  todos
WHERE
  id < 5
  and
  id < 10 -- This does... nothing

This is a redundant condition because it does not change the results! Anything with an ID of less than five necessarily has an ID of less than ten also. You can add or remove this condition, and nothing will change. It's also silly to add because it doesn't provide us any benefit.

We're going to expand our todos table definition a little bit to add due_date and due_time columns. (Storing date and time separately is usually not advised, but it helps us prove the point.)

CREATE TABLE `todos` (
  `id` int NOT NULL AUTO_INCREMENT,
  `title` varchar(255) NOT NULL,
  `due_date` date NOT NULL,
  `due_time` time NOT NULL,
  `created_at` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
  PRIMARY KEY (`id`),
  KEY `due_date` (`due_date`),
  KEY `created_at` (`created_at`)
)

Given this table, if you want to query for todos that are due in the next day, you're stuck using the ADDTIME function:

SELECT
  *
FROM
  todos
WHERE
  ADDTIME(due_date, due_time) BETWEEN NOW() AND NOW() + INTERVAL 1 DAY

We do have an index on due_date, but the index cannot be used because we're performing an operation on it (adding the time). Unlike our previous example, there is no easy way to de-obfuscate this column either since the due_time is different for every row.

We can confirm that the index is not being used by running an EXPLAIN on the previous query:

| id | type | possible_keys | key | key_len | ref | rows  | filtered | Extra       |
|----|------|---------------|-----|---------|-----|-------|----------|-------------|
|  1 | ALL  |               |     |         |     | 39746 |   100.00 | Using where |

To work around this, let's add a redundant condition on due_date alone. When adding the condition, we need to make sure that it's logically impossible to change the result set, which means our redundant condition should be broader than our actual condition.

Since we're looking for todos due in the next 24 hours, we can add a condition that looks for todos due today or tomorrow. That will contain the entire subset of todos that we're looking for and a few that we're not.

EXPLAIN SELECT
  *
FROM
  todos
WHERE
  -- The real condition
  ADDTIME(due_date, due_time) BETWEEN NOW() AND NOW() + INTERVAL 1 DAY
  AND
  -- The redundant condition
  due_date BETWEEN CURRENT_DATE AND CURRENT_DATE + INTERVAL 1 DAY

The redundant condition here returns a broader subset of todos than we need, but importantly it allows MySQL to use the index. Running an EXPLAIN on this query and we see that the due_date index was used:

| id | type  | possible_keys | key      | key_len | ref | rows | filtered | Extra                              |
|----|-------|---------------|----------|---------|-----|------|----------|------------------------------------|
|  1 | range | due_date      | due_date | 3       |     |    1 |   100.00 | Using index condition; Using where |

MySQL will first use the index to eliminate most of the table, then the slower ADDTIME will be used to eliminate the few remaining false positives. The redundant condition is doing its job perfectly!

Domain-specific redundant conditions

Until now, we've been working with redundant conditions that logically cannot change the result set. These are nice because they are easy to reason about and require no further domain knowledge. There are scenarios where you, as a human, might have more knowledge than the database does. (For now, at least.) In those situations, you might be able to add a redundant condition that is not logically incapable of changing the output, but you know, based on your knowledge, that it won't change the output.

In the case of our todos table, let's add an updated_at column that will be populated with the timestamp of the last time the record was changed.

CREATE TABLE `todos` (
  `id` int NOT NULL AUTO_INCREMENT,
  `title` varchar(255) NOT NULL,
  `created_at` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
  `updated_at` timestamp DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP
  PRIMARY KEY (`id`),
  KEY `created_at` (`created_at`)
)

In this scenario, we still only have an index on created_at, but if we want to query against updated_at, we might be able to add a redundant condition based on our knowledge of the application. If, given our understanding of the application, we can be sure that created_at is always equal to or earlier than updated_at, we can use this to our advantage.

This query, which looks for records that were last modified before January 1st of 2023, will scan the entire table because there is no index on updated_at:

SELECT
  *
FROM
  todos
WHERE
  updated_at < '2023-01-01 00:00:00'

This query will return the same results but uses the created_at index to eliminate records and then filters out the false positives.

SELECT
  *
FROM
  todos
WHERE
  updated_at < '2023-01-01 00:00:00'
  AND
  created_at < '2023-01-01 00:00:00'

The only reason this redundant condition works is because we know that a record cannot be modified before it's created. Depending on your application, you might be able to find more examples of "domain-specific" redundant conditions.

When to use a redundant condition

The optimal indexing strategy always depends on the application, but in general, it's best to have indexes on the conditions you are frequently querying against. Redundant conditions are nice because they require no changes to the database! You can modify the query or the application generating the query, and suddenly everything gets faster. This makes them useful for queries that are only sometimes run or where indexes can't be easily added to the main conditions.

If you'd like to learn more about indexing strategies, we have 17 videos on indexes as a part of our larger course on MySQL for Developers.

If you do end up using the redundant condition strategy, please let us know on Twitter how you did it. We'd love to add more examples to this article!