Memcached and Redis Cache

Memcached:lru

Reasons for high concurrency support: multiplexed IO (message notification mode), multithreaded mode (number of cores in CPU)

Distributed scenarios:

# normal hash distribution (modulo)

# Consistent hash distribution (0-232 rings)

Redis

LRU: Remove one of the least recently accessed keys from the database

Random elimination algorithm: delete a key from the database randomly

TTL elimination algorithm: delete one of the fastest expired keys from the database

Object counter: Reference to Counter plus 1, reference Invalid-1, 0 Delete, can avoid concurrency problems.

Type of support:

String:

List: Doubly linked list (implements Message queue, head out, tail put)

Set: Unordered collection (poll, no duplicates)

Sorted set: Ordered set (number of top-post orders)

Hash: Hash table, similar to database tables (save user information)

Transaction processing

Persistence:

#内存快照: Depositing binary files

#日志追加

Master-Slave synchronization: No persistence is allowed on primary, slave can

Step: Slave Active connection master;slave SYCN command to Master request synchronization, master Backup to rdb;master the Rdb file to slave;slave empty the database, the RDB import.

Virtual Memory: When physical memory is low.

Clear data:

#定时清除

#用户获取数据时

Memcached and Redis Cache