Cache

This filter was introduced in MariaDB MaxScale 2.1.

Overview

From MaxScale version 2.2.11 onwards, the cache filter is no longer considered experimental. The following changes to the default behaviour have also been made:

• The default value of cached_data is now thread_specific (used to be shared).
• The default value of selects is now assume_cacheable (used to be verify_cacheable).

The cache filter is a simple cache that is capable of caching the result of SELECTs, so that subsequent identical SELECTs are served directly by MaxScale, without the queries being routed to any server.

By default the cache will be used and populated in the following circumstances:

• There is no explicit transaction active, that is, autocommit is used,
• there is an explicitly read-only transaction (that is,START TRANSACTION READ ONLY) active, or
• there is a transaction active and no statement that modifies the database has been performed.

In practice, the last bullet point basically means that if a transaction has been started with BEGIN, START TRANSACTION or START TRANSACTION READ WRITE, then the cache will be used and populated until the first UPDATE, INSERT or DELETE statement is encountered.

That is, in default mode the cache effectively causes the system to behave as if the isolation level would be READ COMMITTED, irrespective of what the isolation level of the backends actually is.

The default behaviour can be altered using the configuration parameter cache_in_transactions.

By default it is assumed that all SELECT statements are cacheable, which means that also statements like SELECT LOCALTIME are cached. Please check selects for how to change the default behaviour.

Limitations

All of these limitations may be addressed in forthcoming releases.

Prepared Statements

Resultsets of prepared statements are not cached.

Multi-statements

Multi-statements are always sent to the backend and their result is not cached.

Security

The cache is not aware of grants.

The implication is that unless the cache has been explicitly configured who the caching should apply to, the presence of the cache may provide a user with access to data he should not have access to.

Please read the section Security for more detailed information.

However, from 2.5 onwards it is possible to configure the cache to cache the data of each user separately, which effectively means that there can be no unintended sharing. Please see users for how to change the default behaviour.

information_schema

When invalidation is enabled, SELECTs targeting tables in information_schema are not cached. The reason is that as the content of the tables changes as the side-effect of something else, the cache would not know when to invalidate the cache-entries.

Invalidation

Since MaxScale 2.5, the cache is capable of invalidating entries in the cache when a modification (UPDATE, INSERT or DELETE) that may affect those entries is made.

The cache invalidation works on the table-level, that is, a modification made to a particular table will cause all cache entries that refer to that table to be invalidated, irrespective of whether the modification actually has an impact on the cache entries or not. For instance, suppose the result of the following SELECT has been cached

SELECT * FROM t WHERE a=1;

An insert like

INSERT INTO t SET a=42;

will cause the cache entry containing the result of that SELECT to be invalidated even if the INSERT actually does not affect it. Please see invalidate for how to enable the invalidation.

When invalidation has been enabled MaxScale must be able to completely parse a SELECT statement for its results to be stored in the cache. The reason is that in order to be able to invalidate cache entries, MaxScale must know what tables a SELECT statement depends upon. Consequently, if (and only if) invalidation has been enabled and MaxScale fails to parse a statement, the result of that particular statement will not be cached.

When invalidation has been enabled, MaxScale will also parse all UPDATE, INSERT and DELETE statements, in order to find out what tables are modified. If that parsing fails, MaxScale will by default clear the entire cache. The reason is that unless MaxScale can completely parse the statement it cannot know what tables are modified and hence not what cache entries should be invalidated. Consequently, to prevent stale data from being returned, the entire cache is cleared. The default behaviour can be changed using the configuration parameter clear_cache_on_parse_errors.

Note that what threading approach is used has a big impact on the invalidation. Please see Threads, Users and Invalidation for how the threading approach affects the invalidation.

Note also that since the invalidation may not, depending on how the cache has been configured, be visible to all sessions of all users, it is still important to configure a reasonable soft and hard TTL.

Best Efforts

The invalidation offered by the MaxScale cache can be said to be of best efforts quality. The reason is that in order to ensure that the cache in all circumstances reflects the state in the actual database, would require that the operations involving the cache and the MariaDB server are synchronized, which would cause an unacceptable overhead.

What best efforts means in this context is best illustrated using an example.

Suppose a client executes the statement SELECT * FROM tbl and that the result is cached. Next time that or any other client executes the same statement, the result is returned from the cache and the MariaDB server will not be accessed at all.

If a client now executes the statement INSERT INTO tbl VALUES (...), the cached value for the SELECT statement above and all other statements that are dependent upon tbl will be invalidated. That is, the next time someone executes the statement SELECT * FROM tbl the result will again be fetched from the MariaDB server and stored to the cache.

However, suppose some client executes the statement SELECT COUNT(*) FROM tbl at the same time someone else executes the INSERT ... statement. A possible chain of events is as follows:

                      Timeline 1                 Timeline 2

Clients execute       INSERT ...                 SELECT COUNT(*) FROM tbl
MaxScale -> DB                                   SELECT COUNT(*) FROM tbl
MaxScale -> DB        INSERT ...

That is, the SELECT is performed in the database server before the INSERT. However, since the timelines are proceeding independently of each other, the events may be re-ordered as far as the cache is concerned.

MaxScale -> Cache     Delete invalidated values
MaxScale -> Cache                                Store result and invalidation key

That is, the cached value for SELECT COUNT(*) FROM tbl will reflect the situation before the insert and will thus not be correct.

The stale result will be returned until the value has reached its time-to-live or its invalidation is caused by some update operation.

Configuration

The cache is simple to add to any existing service. However, some experimentation may be required in order to find the configuration settings that provide the maximum benefit.

[Cache]
type=filter
module=cache
hard_ttl=30
soft_ttl=20
rules=...
...

[Cached-Routing-Service]
type=service
...
filters=Cache

Each configured cache filter uses a storage of its own. That is, if there are two services, each configured with a specific cache filter, then, even if queries target the very same servers the cached data will not be shared.

Two services can use the same cache filter, but then either the services should use the very same servers or a completely different set of servers, where the used table names are different. Otherwise there can be unintended sharing.

Filter Parameters

The cache filter has no mandatory parameters but a range of optional ones. Note that it is advisable to specify max_size to prevent the cache from using up all memory there is, in case there is very little overlap among the queries.

storage

The name of the module that provides the storage for the cache. That module will be loaded and provided with the value of storage_options as argument. For instance:

storage=storage_inmemory

The default is storage_inmemory.

See Storage for what storage modules are available.

storage_options

A string that is provided verbatim to the storage module specified in storage, when the module is loaded. Note that the needed arguments and their format depend upon the specific module.

hard_ttl

Hard time to live; the maximum amount of time the cached result is used before it is discarded and the result is fetched from the backend (and cached). See also soft_ttl below.

hard_ttl=60s

The default value is 0s, which means no limit.

The duration can be specified as explained here. If no explicit unit has been specified, the value is interpreted as seconds in MaxScale 2.4. In subsequent versions a value without a unit may be rejected.

soft_ttl

Soft time to live; the amount of time - in seconds - the cached result is used before it is refreshed from the server. When soft_ttl has passed, the result will be refreshed when the first client requests the value.

However, as long as hard_ttl has not passed, all other clients requesting the same value will use the result from the cache while it is being fetched from the backend. That is, as long as soft_ttl but not hard_ttl has passed, even if several clients request the same value at the same time, there will be just one request to the backend.

soft_ttl=60s

The default value is 0, which means no limit. If the value of soft_ttl is larger than hard_ttl it will be adjusted down to the same value.

The duration can be specified as explained here. If no explicit unit has been specified, the value is interpreted as seconds in MaxScale 2.4. In subsequent versions a value without a unit may be rejected.

max_resultset_rows

Specifies the maximum number of rows a resultset can have in order to be stored in the cache. A resultset larger than this, will not be stored.

max_resultset_rows=1000

The default value is 0, which means no limit.

max_resultset_size

Specifies the maximum size of a resultset, for it to be stored in the cache. A resultset larger than this, will not be stored. The size can be specified as described here.

max_resultset_size=128Ki

The default value is 0, which means no limit.

Note that the value of max_resultset_size should not be larger than the value of max_size.

max_count

The maximum number of items the cache may contain. If the limit has been reached and a new item should be stored, then an older item will be evicted.

Note that if cached_data is thread_specific then this limit will be applied to each cache separately. That is, if a thread specific cache is used, then the total number of cached items is #threads * the value of max_count.

max_count=1000

The default value is 0, which means no limit.

max_size

The maximum size the cache may occupy. If the limit has been reached and a new item should be stored, then some older item(s) will be evicted to make space. The size can be specified as described here.

Note that if cached_data is thread_specific then this limit will be applied to each cache separately. That is, if a thread specific cache is used, then the total size is #threads * the value of max_size.

max_size=100Mi

The default value is 0, which means no limit.

rules

Specifies the path of the file where the caching rules are stored. A relative path is interpreted relative to the data directory of MariaDB MaxScale.

rules=/path/to/rules-file

cached_data

• Type: enum
• Mandatory: No
• Dynamic: No
• Values: shared, thread_specific
• Default: thread_specific

An enumeration option specifying how data is shared between threads. The allowed values are:

• shared: The cached data is shared between threads. On the one hand it implies that there will be synchronization between threads, on the other hand that all threads will use data fetched by any thread.
• thread_specific: The cached data is specific to a thread. On the one hand it implies that no synchronization is needed between threads, on the other hand that the very same data may be fetched and stored multiple times.

cached_data=shared

Default is thread_specific. See max_count and max_size what implication changing this setting to shared has.

selects

• Type: enum
• Mandatory: No
• Dynamic: No
• Values: assume_cacheable, verify_cacheable
• Default: assume_cacheable

An enumeration option specifying what approach the cache should take with respect to SELECT statements. The allowed values are:

• assume_cacheable: The cache can assume that all SELECT statements, without exceptions, are cacheable.
• verify_cacheable: The cache can not assume that all SELECT statements are cacheable, but must verify that.

selects=verify_cacheable

Default is assume_cacheable. In this case, all SELECT statements are assumed to be cacheable and will be parsed only if some specific rule requires that.

If verify_cacheable is specified, then all SELECT statements will be parsed and only those that are safe for caching - e.g. do not call any non-cacheable functions or access any non-cacheable variables - will be subject to caching.

If verify_cacheable has been specified, the cache will not be used in the following circumstances:

• The SELECT uses any of the following functions: BENCHMARK, CONNECTION_ID, CONVERT_TZ, CURDATE, CURRENT_DATE, CURRENT_TIMESTAMP, CURTIME, DATABASE, ENCRYPT, FOUND_ROWS, GET_LOCK, IS_FREE_LOCK, IS_USED_LOCK, LAST_INSERT_ID, LOAD_FILE, LOCALTIME, LOCALTIMESTAMP, MASTER_POS_WAIT, NOW, RAND, RELEASE_LOCK, SESSION_USER, SLEEP, SYSDATE, SYSTEM_USER, UNIX_TIMESTAMP, USER, UUID, UUID_SHORT.
• The SELECT accesses any of the following fields: CURRENT_DATE, CURRENT_TIMESTAMP, LOCALTIME, LOCALTIMESTAMP
• The SELECT uses system or user variables.

Note that parsing all SELECT statements carries a performance cost. Please read performance for more details.

cache_in_transactions

• Type: enum
• Mandatory: No
• Dynamic: No
• Values: never, read_only_transactions, all_transactions
• Default: all_transactions

An enumeration option specifying how the cache should behave when there are active transactions:

• never: When there is an active transaction, no data will be returned from the cache, but all requests will always be sent to the backend. The cache will be populated inside explicitly read-only transactions. Inside transactions that are not explicitly read-only, the cache will be populated until the first non-SELECT statement.
• read_only_transactions: The cache will be used and populated inside explicitly read-only transactions. Inside transactions that are not explicitly read-only, the cache will be populated, but not used until the first non-SELECT statement.
• all_transactions: The cache will be used and populated inside explicitly read-only transactions. Inside transactions that are not explicitly read-only, the cache will be used and populated until the first non-SELECT statement.

cache_in_transactions=never

Default is all_transactions.

The values read_only_transactions and all_transactions have roughly the same effect as changing the isolation level of the backend to read_committed.

debug

An integer value, using which the level of debug logging made by the cache can be controlled. The value is actually a bitfield with different bits denoting different logging.

• 0 (0b00000) No logging is made.
• 1 (0b00001) A matching rule is logged.
• 2 (0b00010) A non-matching rule is logged.
• 4 (0b00100) A decision to use data from the cache is logged.
• 8 (0b01000) A decision not to use data from the cache is logged.
• 16 (0b10000) Higher level decisions are logged.

Default is 0. To log everything, give debug a value of 31.

debug=31

enabled

Specifies whether the cache is initially enabled or disabled.

enabled=false

Default is true.

The value affects the initial state of the MaxScale user variables using which the behaviour of the cache can be modified at runtime. Please see Runtime Configuration for details.

invalidate

• Type: enum
• Mandatory: No
• Dynamic: No
• Values: never, current
• Default: never

An enumeration option specifying how the cache should invalidate cache entries.

* `never`: No invalidation is performed. This is the default.
* `current`: When a modification is made, entries in the cache used by
  the current session are invalidated. Other sessions that use the same
  cache will also be affected, but sessions that use another cache will
  not.

The effect of current depends upon the value of cached_data. If the value is shared, that is, all threads share the same cache, then the effect of an invalidation is immediately visible to all sessions, as there is just one cache. However, if the value is thread_specific, then an invalidation will affect only the cache that the session happens to be using.

If it is important and sufficient that an application immediately sees a change that it itself has caused, then a combination of invalidate=current and cached_data=thread_specific can be used.

If it is important that an application immediately sees all changes, irrespective of who has caused them, then a combination of invalidate=current and cached_data=shared must be used.

clear_cache_on_parse_errors

This boolean option specifies how the cache should behave in case of parsing errors when invalidation has been enabled.

• true: If the cache fails to parse an UPDATE/INSERT/DELETE statement then all cached data will be cleared.
• false: A failure to parse an UPDATE/INSERT/DELETE statement is ignored and no invalidation will take place due that statement.

The default value is true.

Changing the value to false may mean that stale data is returned from the cache, if an UPDATE/INSERT/DELETE cannot be parsed and the statement affects entries in the cache.

users

• Type: enum
• Mandatory: No
• Dynamic: No
• Values: mixed, isolated
• Default: mixed

An enumeration option specifying how the cache should cache data for different users.

* `mixed`: The data of different users is stored in the same
  cache. This is the default and may cause that a user can
  access data he should not have access to.
* `isolated`: Each user has a unique cache and there can be
  no unintended sharing.

Note that if isolated has been specified, then each user will conceptually have a cache of his own, which is populated independently from each other. That is, if two users make the same query, then the data will be fetched twice and also stored twice. So, a isolated cache will in general use more memory and cause more traffic to the backend compared to a mixed cache.

timeout

The timeout used when performing operations to distributed storages such as redis or memcached.

timeout=7000ms

The default value is 5000ms, that is 5 seconds.

The duration can be specified as explained here.

Runtime Configuration

The cache filter can be configured at runtime by executing SQL commands. If there is more than one cache filter in a service, only the first cache filter will be able to process the variables. The remaining filters will not see them and thus configuring them at runtime is not possible.

@maxscale.cache.populate

Using the variable @maxscale.cache.populate it is possible to specify at runtime whether the cache should be populated or not. Its initial value is the value of the configuration parameter enabled. That is, by default the value is true.

The purpose of this variable is make it possible for an application to decide statement by statement whether the cache should be populated.

SET @maxscale.cache.populate=true;
SELECT a, b FROM tbl;
SET @maxscale.cache.populate=false;
SELECT a, b FROM tbl;

In the example above, the first SELECT will always be sent to the server and the result will be cached, provided the actual cache rules specifies that it should be. The second SELECT may be served from the cache, depending on the value of @maxscale.cache.use (and the cache rules).

The value of @maxscale.cache.populate can be queried

SELECT @maxscale.cache.populate;

but only after it has been explicitly set once.

@maxscale.cache.use

Using the variable @maxscale.cache.use it is possible to specify at runtime whether the cache should be used or not. Its initial value is the value of the configuration parameter enabled. That is, by default the value is true.

The purpose of this variable is make it possible for an application to decide statement by statement whether the cache should be used.

SET @maxscale.cache.use=true;
SELECT a, b FROM tbl;
SET @maxscale.cache.use=false;
SELECT a, b FROM tbl;

The first SELECT will be served from the cache, providing the rules specify that the statement should be cached, the cache indeed contains the result and the date is not stale (as specified by the TTL).

If the data is stale, the SELECT will be sent to the server and the cache entry will be updated, irrespective of the value of @maxscale.cache.populate.

If @maxscale.cache.use is true but the result is not found in the cache, and the result is subsequently fetched from the server, the result will not be added to the cache, unless @maxscale.cache.populate is also true.

The value of @maxscale.cache.use can be queried

SELECT @maxscale.cache.use;

but only after it has explicitly been set once.

@maxscale.cache.soft_ttl

Using the variable @maxscale.cache.soft_ttl it is possible at runtime to specify in seconds what soft ttl should be applied. Its initial value is the value of the configuration parameter soft_ttl. That is, by default the value is 0.

The purpose of this variable is make it possible for an application to decide statement by statement what soft ttl should be applied.

set @maxscale.cache.soft_ttl=600;
SELECT a, b FROM unimportant;
set @maxscale.cache.soft_ttl=60;
SELECT c, d FROM important;

When data is SELECTed from the unimportant table unimportant, the data will be returned from the cache provided it is no older than 10 minutes, but when data is SELECTed from the important table important, the data will be returned from the cache provided it is no older than 1 minute.

Note that @maxscale.cache.hard_ttl overrules @maxscale.cache.soft_ttl in the sense that if the former is less that the latter, then soft ttl will, when used, be adjusted down to the value of hard ttl.

The value of @maxscale.cache.soft_ttl can be queried

SELECT @maxscale.cache.soft_ttl;

but only after it has explicitly been set once.

@maxscale.cache.hard_ttl

Using the variable @maxscale.cache.hard_ttl it is possible at runtime to specify in seconds what hard ttl should be applied. Its initial value is the value of the configuration parameter hard_ttl. That is, by default the value is 0.

The purpose of this variable is make it possible for an application to decide statement by statement what hard ttl should be applied.

Note that as @maxscale.cache.hard_ttl overrules @maxscale.cache.soft_ttl, is is important to ensure that the former is at least as large as the latter and for best overall performance that it is larger.

set @maxscale.cache.soft_ttl=600, @maxscale.cache.hard_ttl=610;
SELECT a, b FROM unimportant;
set @maxscale.cache.soft_ttl=60, @maxscale.cache.hard_ttl=65;
SELECT c, d FROM important;

The value of @maxscale.cache.hard_ttl can be queried

SELECT @maxscale.cache.hard_ttl;

but only after it has explicitly been set once.

Client Driven Caching

With @maxscale.cache.populate and @maxscale.cache.use is it possible to make the caching completely client driven.

Provide no rules file, which means that all SELECT statements are subject to caching and that all users receive data from the cache. Set the startup mode of the cache to disabled.

[TheCache]
type=filter
module=cache
enabled=false

Now, in order to mark statements that should be cached, set @maxscale.cache.populate to true, and perform those SELECTs.

SET @maxscale.cache.populate=true;
SELECT a, b FROM tbl1;
SELECT c, d FROM tbl2;
SELECT e, f FROM tbl3;
SET @maxscale.cache.populate=false;

Note that those SELECTs must return something in order for the statement to be marked for caching.

After this, the value of @maxscale.cache.use will decide whether or not the cache is considered.

SET @maxscale.cache.use=true;
SELECT a, b FROM tbl1;
SET @maxscale.cache.use=false;

With @maxscale.cache.use being true, the cache is considered and the result returned from there, if not stale. If it is stale, the result is fetched from the server and the cached entry is updated.

By setting a very long TTL it is possible to prevent the cache from ever considering an entry to be stale and instead manually cause the cache to be updated when needed.

UPDATE tbl1 SET a = ...;
SET @maxscale.cache.populate=true;
SELECT a, b FROM tbl1;
SET @maxscale.cache.populate=false;

Threads, Users and Invalidation

What caching approach is used and how different users are treated has a significant impact on the behaviour of the cache. In the following the implication of different combinations is explained.

Invalidation

Invalidation takes place only in the current cache, so how visible the invalidation is, depends upon the configuration value of cached_data.

cached_data=thread_specific

The invalidation is visible only to the sessions that are handled by the same worker thread where the invalidation occurred. Sessions of the same or other users that are handled by different worker threads will not see the new value before the TTL causes the value to be refreshed.

cache_data=shared

The invalidation is immediately visible to all sessions of all users.

Rules

The caching rules are expressed as a JSON object or as an array of JSON objects.

There are two decisions to be made regarding the caching; in what circumstances should data be stored to the cache and in what circumstances should the data in the cache be used.

Expressed in JSON this looks as follows

{
    store: [ ... ],
    use: [ ... ]
}

or, in case an array is used, as

[
    {
        store: [ ... ],
        use: [ ... ]
    },
    { ... }
]

The store field specifies in what circumstances data should be stored to the cache and the use field specifies in what circumstances the data in the cache should be used. In both cases, the value is a JSON array containing objects.

If an array of rule objects is specified, then, when looking for a rule that matches, the store field of each object are evaluated in sequential order until a match is found. Then, the use field of that object is used when deciding whether data in the cache should be used.

When to Store

By default, if no rules file have been provided or if the store field is missing from the object, the results of all queries will be stored to the cache, subject to max_resultset_rows and max_resultset_size cache filter parameters.

By providing a store field in the JSON object, the decision whether to store the result of a particular query to the cache can be controlled in a more detailed manner. The decision to cache the results of a query can depend upon

• the database,
• the table,
• the column, or
• the query itself.

Each entry in the store array is an object containing three fields,

{
    "attribute": <string>,
    "op": <string>
    "value": <string>
}

where,

• the attribute can be database, table, column or query,
• the op can be =, !=, like or unlike, and
• the value a string.

If op is = or != then value is used as a string; if it is like or unlike, then value is interpreted as a pcre2 regular expression. Note though that if attribute is database, table or column, then the string is interpreted as a name, where a dot . denotes qualification or scoping.

The objects in the store array are processed in order. If the result of a comparison is true, no further processing will be made and the result of the query in question will be stored to the cache.

If the result of the comparison is false, then the next object is processed. The process continues until the array is exhausted. If there is no match, then the result of the query is not stored to the cache.

Note that as the query itself is used as the key, although the following queries

select * from db1.tbl

and

use db1;
select * from tbl

target the same table and produce the same results, they will be cached separately. The same holds for queries like

select * from tbl where a = 2 and b = 3;

and

select * from tbl where b = 3 and a = 2;

as well. Although they conceptually are identical, there will be two cache entries.

Note that if a column has been specified in a rule, then a statement will match irrespective of where that particular column appears. For instance, if a rule specifies that the result of statements referring to the the column a should be cached, then the following statement will match

select a from tbl;

and so will

select b from tbl where a > 5;

Qualified Names

When using = or != in the rule object in conjunction with database, table and column, the provided string is interpreted as a name, that is, dot (.) denotes qualification or scope.

In practice that means that if attribute is database then value may not contain a dot, if attribute is table then value may contain one dot, used for separating the database and table names respectively, and if attribute is column then value may contain one or two dots, used for separating table and column names, or database, table and column names.

Note that if a qualified name is used as a value, then all parts of the name must be available for a match. Currently Maria DB MaxScale may not always be capable of deducing in what table a particular column is. If that is the case, then a value like tbl.field may not necessarily be a match even if the field is field and the table actually is tbl.

Implication of the default database

If the rules concerns the database, then only if the statement refers to no specific database, will the default database be considered.

Regexp Matching

The string used for matching the regular expression contains as much information as there is available. For instance, in a situation like

use somedb;
select fld from tbl;

the string matched against the regular expression will be somedb.tbl.fld.

Examples

Cache all queries targeting a particular database.

{
    "store": [
        {
            "attribute": "database",
            "op": "=",
            "value": "db1"
        }
    ]
}

Cache all queries not targeting a particular table

{
    "store": [
        {
            "attribute": "table",
            "op": "!=",
            "value": "tbl1"
        }
    ]
}

That will exclude queries targeting table tbl1 irrespective of which database it is in. To exclude a table in a particular database, specify the table name using a qualified name.

{
    "store": [
        {
            "attribute": "table",
            "op": "!=",
            "value": "db1.tbl1"
        }
    ]
}

Cache all queries containing a WHERE clause

{
    "store": [
        {
            "attribute": "query",
            "op": "like",
            "value": ".*WHERE.*"
        }
    ]
}

Note that that will actually cause all queries that contain WHERE anywhere, to be cached.

When to Use

By default, if no rules file have been provided or if the use field is missing from the object, all users may be returned data from the cache.

By providing a use field in the JSON object, the decision whether to use data from the cache can be controlled in a more detailed manner. The decision to use data from the cache can depend upon

• the user.

Each entry in the use array is an object containing three fields,

{
    "attribute": <string>,
    "op": <string>
    "value": <string>
}

where,

• the attribute can be user,
• the op can be =, !=, like or unlike, and
• the value a string.

If op is = or != then value is interpreted as a MariaDB account string, that is, % means indicates wildcard, but if op is like or unlike it is simply assumed value is a pcre2 regular expression.

For instance, the following are equivalent:

{
    "attribute": "user",
    "op": "=",
    "value": "'bob'@'%'"
}

{
    "attribute": "user",
    "op": "like",
    "value": "bob@.*"
}

Note that if op is = or != then the usual assumptions apply, that is, a value of bob is equivalent with 'bob'@'%'. If like or unlike is used, then no assumptions apply, but the string is used verbatim as a regular expression.

The objects in the use array are processed in order. If the result of a comparison is true, no further processing will be made and the data in the cache will be used, subject to the value of ttl.

If the result of the comparison is false, then the next object is processed. The process continues until the array is exhausted. If there is no match, then data in the cache will not be used.

Note that use is relevant only if the query is subject to caching, that is, if all queries are cached or if a query matches a particular rule in the store array.

Examples

Use data from the cache for all users except admin (actually 'admin'@'%'), regardless of what host the admin user comes from.

{
    "use": [
        {
            "attribute": "user",
            "op": "!=",
            "value": "admin"
        }
    ]
}

Security

As the cache is not aware of grants, unless the cache has been explicitly configured who the caching should apply to, the presence of the cache may provide a user with access to data he should not have access to. Note that the following applies only if users=mixed has been configured. If users=isolated has been configured, then there can never be any unintended sharing between users.

Suppose there is a table access that the user alice has access to, but the user bob does not. If bob tries to access the table, he will get an error as reply:

MySQL [testdb]> select * from access;
ERROR 1142 (42000): SELECT command denied to user 'bob'@'localhost' for table 'access'

If we now setup caching for the table, using the simplest possible rules file, bob will get access to data from the table, provided he executes a select identical with one alice has executed.

For instance, suppose the rules look as follows:

{
    "store": [
        {
            "attribute": "table",
            "op": "=",
            "value": "access"
        }
    ]
}

If alice now queries the table, she will get the result, which also will be cached:

MySQL [testdb]> select * from access;
+------+------+
| a    | b    |
+------+------+
|   47 |   11 |
+------+------+

If bob now executes the very same query, and the result is still in the cache, it will be returned to him.

MySQL [testdb]> select current_user();
+----------------+
| current_user() |
+----------------+
| bob@127.0.0.1  |
+----------------+
1 row in set (0.00 sec)

MySQL [testdb]> select * from access;
+------+------+
| a    | b    |
+------+------+
|   47 |   11 |
+------+------+

That can be prevented, by explicitly declaring in the rules that the caching should be applied to alice only.

{
    "store": [
        {
            "attribute": "table",
            "op": "=",
            "value": "access"
        }
    ],
    "use": [
        {
            "attribute": "user",
            "op": "=",
            "value": "'alice'@'%'"
        }
    ]
}

With these rules in place, bob is again denied access, since queries targeting the table access will in his case not be served from the cache.

Storage

There are two types of storages that can be used; local and shared.

The only local storage implementation is storage_inmemory that simply stores the cache values in memory. The storage is not persistent and is destroyed when MaxScale terminates. Since the storage exists in the MaxScale process, it is very fast and provides almost always a performance benefit.

Currently there are two shared storages; storage_memcached and storage_redis that are implemented using memcached and redis respectively.

The shared storages are accessed across the network and consequently it is not self-evident that their use will provide any performance benefit. Namely, irrespective of whether the data is fetched from the cache or from the server there will be a network hop and often that network hop is, as far as the performance goes, what costs the most.

The presence of a shared cache may provide a performance benefit if the network between MaxScale and the storage server (memcached or Redis) is faster than the network between MaxScale and the database server, if the used SELECT statements are heavy (that is, take a significant amount of time) to process for the database server, or * if the presence of the cache reduces the overall load of an otherwise overloaded database server.

As a general rule a shared storage should not be used without first assessing its value using a realistic workload.

storage_inmemory

This simple storage module uses the standard memory allocator for storing the cached data.

storage=storage_inmemory

This storage module takes no arguments.

storage_memcached

This storage module uses memcached for storing the cached data.

Multiple MaxScale instances can share the same memcached server and items cached by one MaxScale instance will be used by the other. Note that all MaxScale instances should have exactly the same configuration, as otherwise there can be unintended sharing.

storage=storage_memcached

storage_memcache has the following mandatory arguments:

• server using which the location of the server is specified as host[:port]. If no port is provided, the default Memcached port of 11211 is used.

storage_memcached has the following optional arguments:

• max_value_size using which the maximum size of a cached value is specified. By default, the maximum size of a value stored to memcached is 1MB, but that configured to be something else. The value of max_value_size will be used for capping max_resultset_size, that is, unless memcached is configured to allow larger values that 1M and max_value_size has been set accordingly, only resultsets up to 1MB in size will be cached. The value can be specified as documented here.

Example:

storage_options="server=192.168.1.31:11211, max_value_size=10M"

Limitations

• Invalidation is not supported.
• Configuration values given to max_size and max_count are ignored.

Security

Neither the data in the memcached server nor the traffic between MaxScale and the memcached server is encrypted. Consequently, anybody with access to the memcached server or to the network have access to the cached data.

storage_redis

This storage module uses redis for storing the cached data.

Multiple MaxScale instances can share the same redis server and items cached by one MaxScale instance will be used by the other. Note that all MaxScale instances should have exactly the same configuration, as otherwise there can be unintended sharing.

storage=storage_redis

If storage_redis cannot connect to the Redis server, caching will silently be disabled and a connection attempt will be made after a timeout interval.

If a timeout error occurs during an operation, reconnecting will be attempted after a delay, which will be an increasing multiple of timeout. For example, if timeout is the default 5 seconds, then reconnection attempts will first be made after 10 seconds, then after 15 seconds, then 20 and so on. However, once 60 seconds have been reached, the delay will no longer be increased but the delay will stay at one minute. Note that each time a reconnection attempt is made, unless the reason for the timeout has disappeared, the client will be stalled for timeout seconds.

storage_redis has the following mandatory arguments:

• server using which the location of the server is specified as host[:port]. If no port is provided, the default Redis port of 6379 is used.

Example:

storage_options="server=192.168.1.31:6379"

Note that Redis should be configured with no idle timeout or with a timeout that is very large. Otherwise MaxScale may have to repeatedly connect to Redis, which will hurt both the functionality and the performance.

Limitations

• There is no distinction between soft and hard ttl, but only hard ttl is used.
• Configuration values given to max_size and max_count are ignored.

Invalidation

storage_redis supports invalidation, but the caveats documented here are of greater significance since also the communication between the cache and the cache storage is asynchronous and takes place over the network.

NOTE If invalidation is turned on after caching has been used (in non-invalidation mode), redis must be flushed as otherwise there will be entries in the cache that will not be affected by the invalidation.

$ redis-cli flushall

Security

Neither the data in the redis server nor the traffic between MaxScale and the redis server is encrypted. Consequently, anybody with access to the redis server or to the network have access to the cached data.

Example

In the following we define a cache MyCache that uses the cache storage module storage_inmemory and whose soft ttl is 30 seconds and whose hard ttl is 45 seconds. The cached data is shared between all threads and the maximum size of the cached data is 50 mebibytes. The rules for the cache are in the file cache_rules.json.

Configuration

[MyCache]
type=filter
module=cache
storage=storage_inmemory
soft_ttl=30
hard_ttl=45
cached_data=shared
max_size=50Mi
rules=cache_rules.json

[MyService]
type=service
...
filters=MyCache

cache_rules.json

The rules specify that the data of the table sbtest should be cached.

{
    "store": [
        {
            "attribute": "table",
            "op": "=",
            "value": "sbtest"
        }
    ]
}

Performance

When the cache filter was introduced, the most significant factor affecting the performance of the cache was whether the statements needed to be parsed. Initially, all statements were parsed in order to exclude SELECT statements that use non-cacheable functions, access non-cacheable variables or refer to system or user variables. Later, the default value of the selects parameter was changed to assume_cacheable, to maximize the default performance.

With the default configuration, the cache itself will not cause the statements to be parsed. However, even with assume_cacheable configured, a rule referring specifically to a database, table or column will still cause the statement to be parsed.

For instance, a simple rule like

{
    "store": [
        {
            "attribute": "database",
            "op": "=",
            "value": "db1"
        }
    ]
}

cannot be fulfilled without parsing the statement.

If the rule is instead expressed using a regular expression

{
    "store": [
        {
            "attribute": "query",
            "op": "like",
            "value": "FROM db1\\..*"
        }
    ]
}

then the statement will not be parsed.

However, when the query classifier cache was introduced, the parsing cost was significantly reduced and currently the cost for parsing and regular expression matching is roughly the same.

In the following is a table with numbers giving a rough picture of the relative cost of different approaches.

In the table, regexp match means that the cacheable statements were picked out using a rule like

{
    "attribute": "query",
    "op": "unlike",
    "value": "FROM nomatch"
}

while exact match means that the cacheable statements were picked out using a rule like

{
    "attribute": "database",
    "op": "!=",
    "value": "nomatch"
}

The exact match rule requires all statements to be parsed.

As the purpose of the test is to illustrate the overhead of different approaches, the rules were formulated so that all SELECT statements would match.

Note that these figures were obtained by running sysbench, MaxScale and the server in the same computer, so they are only indicative.

For comparison, without caching, the the qps is 33.

As can be seen, due to the query classifier cache there is no difference between exact and regex based matching.

Summary

For maximum performance:

• Arrange the situation so that the default selects=assume_cacheable can be used, and use no rules.

Otherwise it is mostly a personal preference whether exact or regex based rules are used. However, one should always test with real data and real queries before choosing one over the other.