Redis系列20：LFU内存淘汰算法分析

     16 bits      8 bits
+----------------+--------+
+ Last decr time | LOG_C  |
+----------------+--------+
1.
2.
3.
4.

/* Return the current time in minutes, just taking the least significant
 * 16 bits. The returned time is suitable to be stored as LDT (last decrement
 * time) for the LFU implementation. */
// server.unixtime为Redis缓存的Unix时间戳
// 使用的Unix的分钟时间戳，取模2^16
unsigned long LFUGetTimeInMinutes(void) {
  return (server.unixtime/60) & 65535;
}
 
/* Given an object last access time, compute the minimum number of minutes
 * that elapsed since the last access. Handle overflow (ldt greater than
 * the current 16 bits minutes time) considering the time as wrapping
 * exactly once. */
unsigned long LFUTimeElapsed(unsigned long ldt) {
  // 获取系统当前的LFU time
  unsigned long now = LFUGetTimeInMinutes();
  // 如果now >= ldt 直接取差值  
  if (now >= ldt) return now-ldt;
  // 如果now < ldt 增加上65535
  return 65535-ldt+now;
}
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.

/* Logarithmically increment a counter. The greater is the current counter value
 * the less likely is that it gets really implemented. Saturate it at 255. */
uint8_t LFULogIncr(uint8_t counter) {
  // Logistic Counter最大值为255 （8位的最大值）,  如果已经是最大值了，直接返回
  if (counter == 255) return 255;
  // 取一个0~1之间的随机数数
  double r = (double)rand()/RAND_MAX;
  // counter减去LFU_INIT_VAL （LFU_INIT_VAL为每个key的Logistic Counter基数值，默认为5）
  double baseval = counter - LFU_INIT_VAL;
  // 如果衰减之后counter已经小于基数（如5），那么得出的结果 < 0，也取0
  if (baseval < 0) baseval = 0;
  // 可以看出如果lfu_log_factor的值越大，分母越大，得到的p越小
  double p = 1.0/(baseval*server.lfu_log_factor+1);
    // p 越小，r < p的可能性就越小，Logistic Counter增加的概率就越小
	// 综上，lfu_log_factor越大增长越缓慢，缓解255空间紧张的问题
  if (r < p) counter++;
  return counter;
}
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.

# MAXMEMORY POLICY: how Redis will select what to remove when maxmemory
# is reached. You can select one from the following behaviors:
#
# volatile-lru -> Evict using approximated LRU, only keys with an expire set.
# allkeys-lru -> Evict any key using approximated LRU.
# volatile-lfu -> Evict using approximated LFU, only keys with an expire set.
# allkeys-lfu -> Evict any key using approximated LFU.
# volatile-random -> Remove a random key having an expire set.
# allkeys-random -> Remove a random key, any key.
# volatile-ttl -> Remove the key with the nearest expire time (minor TTL)
# noeviction -> Don't evict anything, just return an error on write operations.
#
# LRU means Least Recently Used
# LFU means Least Frequently Used
#
# Both LRU, LFU and volatile-ttl are implemented using approximated
# randomized algorithms.
#
# Note: with any of the above policies, when there are no suitable keys for
# eviction, Redis will return an error on write operations that require
# more memory. These are usually commands that create new keys, add data or
# modify existing keys. A few examples are: SET, INCR, HSET, LPUSH, SUNIONSTORE,
# SORT (due to the STORE argument), and EXEC (if the transaction includes any
# command that requires memory).
#
# The default is:
#
# maxmemory-policy noeviction
#
#
# 备注1：对设置了过期时间的key启用LFU淘汰算法
# maxmemory-policy volatile-lfu
# 备注2：对全部key启用LFU淘汰算法进行计算
# maxmemory-policy allkeys-lfu
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.