How to solve the problem when redis memory is full

Redis memory is full Solution:

1, increase the memory.

2, use memory elimination strategy.

3, Redis cluster.

Focus on 2 and 3:

The second point:

We know that the maxmemory parameter of the redis configuration file can control its maximum available memory size (bytes ).

So what should I do when the required memory exceeds maxmemory?

At this time, the maxmemory-policy in the configuration file comes into play.

The default value is noeviction.

Below I will list the elimination rules for deleting redis keys when the available memory is insufficient.

Rule description:

1. volatile-lru

Use the LRU algorithm to delete a key (only for keys with a set survival time)

2. allkeys-lru

Use the LRU algorithm to delete a key

3, volatile-random

Delete a key randomly (only for keys with a set survival time)

4. allkeys-random

Delete a key randomly

5.volatile-ttl

Delete the key with the latest survival time

6.noeviction

Do not delete the key, only return an error

LRU algorithm, least recently used, least recently used algorithm. That is to say, the least recently used key is deleted by default.

But you must pay attention to one thing! Redis does not accurately delete the least recently used key among all keys, but randomly selects 3 keys and deletes the least recently used key among these three keys.

Then the number 3 can also be set, and the corresponding location is maxmeory-samples in the configuration file.

3. How to do the cluster

Redis only supports single instance, memory Generally up to 10~20GB. For systems with 100 to 200GB of memory, it needs to be supported through clustering.

There are three methods of Redis clustering: client sharding, proxy sharding, and RedisCluster (more on this in a later article.)

1. Client sharding

Implement routing through business code

Advantages: You can control the sharding algorithm yourself, and the performance is better than that of the proxy

Disadvantages: High maintenance costs, expansion/shrinking and other operation and maintenance operations need to be done by yourself Research and development

2. Agent sharding

The agent program receives data requests from the business program, and distributes these requests to the correct Redis instance according to the routing rules and returns them to the business program. Implemented using middleware such as Twemproxy and Codis.

Advantages: Easy operation and maintenance, the program does not need to worry about how to connect the Redis instance

Disadvantages: It will cause performance consumption (about 20%), cannot smoothly expand/shrink, and needs to execute scripts to migrate data , inconvenient (Codis optimizes and implements pre-sharding based on Twemproxy to achieve Auto Rebalance).

3. Redis Cluster

Advantages: Official cluster solution, no central node, direct connection to the client, good performance

Disadvantages: The solution is too heavy and cannot be smoothed Expansion/reduction requires the execution of corresponding scripts, which is inconvenient, too new, and there are no corresponding mature solutions.

For more Redis-related knowledge, please visit the Redis Usage Tutorial column!

The above is the detailed content of How to solve the problem when redis memory is full. For more information, please follow other related articles on the PHP Chinese website!