Redis系列之淘汰策略介绍

为什么需要Redis淘汰策略？

由于Redis内存是有大小的，当内存快满的时候，又没有过期数据，这个时候就会导致内存被占满，内存满了，自然就不能再放入新的数据。所以，就需要Redis的淘汰策略来保证可用性。

Redis淘汰策略分类

在Redis中提供了好几种淘汰策略，查看官方文档

https://redis.io/docs/latest/operate/rs/databases/memory-performance/eviction-policy/，找到如下几种淘汰策略：

Eviction Policy	Description
noeviction	New values aren’t saved when memory limit is reached When a database uses replication, this applies to the primary database // 默认策略，默认不淘汰数据，能读不能写
allkeys-lru	Keeps most recently used keys; removes least recently used (LRU) keys // 基于伪LRU算法，在所有的key中去淘汰
allkeys-lfu	Keeps frequently used keys; removes least frequently used (LFU) keys // 基于伪LRU算法，在所有的key中去淘汰
allkeys-random	Randomly removes keys // 基于随机算法，在所有的key中去淘汰
volatile-lru	Removes least recently used keys with expire field set to true // 基于伪LRU算法，在设置了过期时间的key中去淘汰
volatile-lfu	Removes least frequently used keys with expire field set to true // 基于伪LFU算法，在设置了过期时间的key中去淘汰
volatile-random	Randomly removes keys with expire field set to true // 基于随机算法，在设置了过期时间的key中去淘汰
volatile-ttl	Removes least frequently used keys with expire field set to true and the shortest remaining time-to-live (TTL) value // 根据过期时间来，淘汰即将过期的

我们发现redis提供了8种不同的策略，只要在我们的config中配置maxmemory-policy即可指定相关的淘汰策略。

maxmemory-policy noeviction # 默认淘汰策略，只能读不能写

Redis数据淘汰流程

淘汰流程：

1. 首先，我们会有一个淘汰池，默认大小是16，并且里面的数据都是末尾淘汰机制。
2. 每次指令操作的时候，会自旋判断当前的内存是否满足指令所需要的内存，内存满足，继续指令操作
3. 如果当前内存不能满足时，判断淘汰机制是否为noeviction，是默认的noeviction机制，OOM报错给用户，只能读不能写，如果不是默认的noeviction机制会从淘汰池中的尾部拿取一个最适合淘汰的数据。
   1. 取样，从Redis中随机获取取样的数据，不一次性读取所有的数据。
   2. 在取样的数据中，根据淘汰算法，找到最适合淘汰的数据
   3. 将最合适淘汰的取样数据跟淘汰池中的数据比较，是否比淘汰池中的数据更适合淘汰，如果更合适，才放入淘汰池
   4. 淘汰池按照适合的程度进行排序，最适合的数据放在尾部
4. 将需要淘汰的数据从redis中删除，并且从淘汰池移除

源码验证淘汰流程

每次执行操作指令都会走freeMemoryIfNeeded函数(evict.c文件)

/* This function is periodically called to see if there is memory to free
 * according to the current "maxmemory" settings. In case we are over the
 * memory limit, the function will try to free some memory to return back
 * under the limit.
 *
 * The function returns C_OK if we are under the memory limit or if we
 * were over the limit, but the attempt to free memory was successful.
 * Otherwise if we are over the memory limit, but not enough memory
 * was freed to return back under the limit, the function returns C_ERR. */
int freeMemoryIfNeeded(void) {
    int keys_freed = 0;
    /* By default replicas should ignore maxmemory
     * and just be masters exact copies. */
    /* 从库是否忽略内存淘汰机制，server.masterhost有配置，说明是从库 */
    if (server.masterhost && server.repl_slave_ignore_maxmemory) return C_OK;

    size_t mem_reported, mem_tofree, mem_freed;
    mstime_t latency, eviction_latency, lazyfree_latency;
    long long delta;
    int slaves = listLength(server.slaves);
    int result = C_ERR;

    /* When clients are paused the dataset should be static not just from the
     * POV of clients not being able to write, but also from the POV of
     * expires and evictions of keys not being performed. */
    if (clientsArePaused()) return C_OK;
    /* 判断内存是否满，如果没有超过内存，直接返回 */
    if (getMaxmemoryState(&mem_reported,NULL,&mem_tofree,NULL) == C_OK)
        return C_OK;

    mem_freed = 0;

    latencyStartMonitor(latency);
    
    /* 如果策略为noeviction，默认不淘汰数据，直接报错OOM */
    if (server.maxmemory_policy == MAXMEMORY_NO_EVICTION)
        goto cant_free; /* We need to free memory, but policy forbids. */
	
    /* 内存不够的情况，一直自旋释放内存 */
    while (mem_freed < mem_tofree) {
        int j, k, i;
        static unsigned int next_db = 0;
        sds bestkey = NULL; // 定义最好的删除key
        int bestdbid;
        redisDb *db;
        dict *dict;
        dictEntry *de;
	
        if (server.maxmemory_policy & (MAXMEMORY_FLAG_LRU|MAXMEMORY_FLAG_LFU) ||
            server.maxmemory_policy == MAXMEMORY_VOLATILE_TTL)
        { // 如果淘汰算法是LRU、LFU、TTL
            struct evictionPoolEntry *pool = EvictionPoolLRU; // 淘汰池，默认大小为16
			// 自旋，找到合适的要淘汰的key
            while(bestkey == NULL) {
                unsigned long total_keys = 0, keys;

                /* We don't want to make local-db choices when expiring keys,
                 * so to start populate the eviction pool sampling keys from
                 * every DB. */
                 /* 去不同的DB查找 */
                for (i = 0; i < server.dbnum; i++) {
                    db = server.db+i;
                    dict = (server.maxmemory_policy & MAXMEMORY_FLAG_ALLKEYS) ?
                            db->dict : db->expires; // 判断需要淘汰的范围，是所有数据还是过期的数据
                    if ((keys = dictSize(dict)) != 0) { 
                        evictionPoolPopulate(i, dict, db->dict, pool);// 关键方法，从范围中取样，拿到最适合淘汰的数据
                        total_keys += keys;
                    }
                }
                if (!total_keys) break; /* No keys to evict. */ /*没有过期的key*/

                /* Go backward from best to worst element to evict. */
                for (k = EVPOOL_SIZE-1; k >= 0; k--) {
                    if (pool[k].key == NULL) continue;
                    bestdbid = pool[k].dbid;

                    if (server.maxmemory_policy & MAXMEMORY_FLAG_ALLKEYS) {
                        de = dictFind(server.db[pool[k].dbid].dict,
                            pool[k].key);
                    } else {
                        de = dictFind(server.db[pool[k].dbid].expires,
                            pool[k].key);
                    }

                    /* Remove the entry from the pool. */
                    if (pool[k].key != pool[k].cached)
                        sdsfree(pool[k].key);
                    pool[k].key = NULL;
                    pool[k].idle = 0;

                    /* If the key exists, is our pick. Otherwise it is
                     * a ghost and we need to try the next element. */
                    if (de) {
                        bestkey = dictGetKey(de);
                        break;
                    } else {
                        /* Ghost... Iterate again. */
                    }
                }
            }
        }

        /* volatile-random and allkeys-random policy */
        else if (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM ||
                 server.maxmemory_policy == MAXMEMORY_VOLATILE_RANDOM)
        {
            /* When evicting a random key, we try to evict a key for
             * each DB, so we use the static 'next_db' variable to
             * incrementally visit all DBs. */
            for (i = 0; i < server.dbnum; i++) {
                j = (++next_db) % server.dbnum;
                db = server.db+j;
                dict = (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM) ?
                        db->dict : db->expires;
                if (dictSize(dict) != 0) {
                    de = dictGetRandomKey(dict);
                    bestkey = dictGetKey(de);
                    bestdbid = j;
                    break;
                }
            }
        }

        /* Finally remove the selected key. */
        /* 移除这个key */
        if (bestkey) {
            db = server.db+bestdbid;
            robj *keyobj = createStringObject(bestkey,sdslen(bestkey));
            propagateExpire(db,keyobj,server.lazyfree_lazy_eviction);
            /* We compute the amount of memory freed by db*Delete() alone.
             * It is possible that actually the memory needed to propagate
             * the DEL in AOF and replication link is greater than the one
             * we are freeing removing the key, but we can't account for
             * that otherwise we would never exit the loop.
             *
             * Same for CSC invalidation messages generated by signalModifiedKey.
             *
             * AOF and Output buffer memory will be freed eventually so
             * we only care about memory used by the key space. */
            delta = (long long) zmalloc_used_memory();
            latencyStartMonitor(eviction_latency);
            /* 如果是异步淘汰，进行异步淘汰*/
            if (server.lazyfree_lazy_eviction)
                dbAsyncDelete(db,keyobj);// 异步淘汰机制
            else
                dbSyncDelete(db,keyobj); // 同步淘汰机制
            latencyEndMonitor(eviction_latency);
            latencyAddSampleIfNeeded("eviction-del",eviction_latency);
            delta -= (long long) zmalloc_used_memory();
            mem_freed += delta;
            server.stat_evictedkeys++;
            signalModifiedKey(NULL,db,keyobj);
            notifyKeyspaceEvent(NOTIFY_EVICTED, "evicted",
                keyobj, db->id);
            decrRefCount(keyobj);
            keys_freed++;

            /* When the memory to free starts to be big enough, we may
             * start spending so much time here that is impossible to
             * deliver data to the slaves fast enough, so we force the
             * transmission here inside the loop. */
            if (slaves) flushSlavesOutputBuffers();

            /* Normally our stop condition is the ability to release
             * a fixed, pre-computed amount of memory. However when we
             * are deleting objects in another thread, it's better to
             * check, from time to time, if we already reached our target
             * memory, since the "mem_freed" amount is computed only
             * across the dbAsyncDelete() call, while the thread can
             * release the memory all the time. */
            if (server.lazyfree_lazy_eviction && !(keys_freed % 16)) {
                if (getMaxmemoryState(NULL,NULL,NULL,NULL) == C_OK) {
                    /* Let's satisfy our stop condition. */
                    mem_freed = mem_tofree;
                }
            }
        } else {
            goto cant_free; /* nothing to free... */
        }
    }
    result = C_OK;

cant_free:
    /* We are here if we are not able to reclaim memory. There is only one
     * last thing we can try: check if the lazyfree thread has jobs in queue
     * and wait... */
    if (result != C_OK) {
        latencyStartMonitor(lazyfree_latency);
        while(bioPendingJobsOfType(BIO_LAZY_FREE)) {
            if (getMaxmemoryState(NULL,NULL,NULL,NULL) == C_OK) {
                result = C_OK;
                break;
            }
            usleep(1000);
        }
        latencyEndMonitor(lazyfree_latency);
        latencyAddSampleIfNeeded("eviction-lazyfree",lazyfree_latency);
    }
    latencyEndMonitor(latency);
    latencyAddSampleIfNeeded("eviction-cycle",latency);
    return result;
}

evictionPoolPopulate方法(evict.c文件)

/* This is an helper function for freeMemoryIfNeeded(), it is used in order
 * to populate the evictionPool with a few entries every time we want to
 * expire a key. Keys with idle time smaller than one of the current
 * keys are added. Keys are always added if there are free entries.
 *
 * We insert keys on place in ascending order, so keys with the smaller
 * idle time are on the left, and keys with the higher idle time on the
 * right. */

void evictionPoolPopulate(int dbid, dict *sampledict, dict *keydict, struct evictionPoolEntry *pool) {
    int j, k, count;
    // 需要取样的数据
    dictEntry *samples[server.maxmemory_samples];
	// 随机从需要取样的范围中得到取样的数据
    count = dictGetSomeKeys(sampledict,samples,server.maxmemory_samples);
    // 循环取样数据
    for (j = 0; j < count; j++) {
        unsigned long long idle;
        sds key;
        robj *o;
        dictEntry *de;

        de = samples[j];
        key = dictGetKey(de);

        /* If the dictionary we are sampling from is not the main
         * dictionary (but the expires one) we need to lookup the key
         * again in the key dictionary to obtain the value object. */
        if (server.maxmemory_policy != MAXMEMORY_VOLATILE_TTL) { // 如果是ttl，只能从带有过期时间的数据中获取，所以不需要获取对象，其它的淘汰策略都需要去我们的键值对中获取值对象
            if (sampledict != keydict) de = dictFind(keydict, key);
            o = dictGetVal(de);
        }

        /* Calculate the idle time according to the policy. This is called
         * idle just because the code initially handled LRU, but is in fact
         * just a score where an higher score means better candidate. */
        if (server.maxmemory_policy & MAXMEMORY_FLAG_LRU) { // 如果是LRU算法，采用LRU算法得到最长时间没访问的
            idle = estimateObjectIdleTime(o);
        } else if (server.maxmemory_policy & MAXMEMORY_FLAG_LFU) { // 如果是LFU算法，根据LFU算法得到最少访问的，idle越大，越容易淘汰，因为是用255-LFUDecrAndReturn(o);
            /* When we use an LRU policy, we sort the keys by idle time
             * so that we expire keys starting from greater idle time.
             * However when the policy is an LFU one, we have a frequency
             * estimation, and we want to evict keys with lower frequency
             * first. So inside the pool we put objects using the inverted
             * frequency subtracting the actual frequency to the maximum
             * frequency of 255. */
            idle = 255-LFUDecrAndReturn(o);
        } else if (server.maxmemory_policy == MAXMEMORY_VOLATILE_TTL) { // ttl 直接根据时间来
            /* In this case the sooner the expire the better. */
            idle = ULLONG_MAX - (long)dictGetVal(de);
        } else {
            serverPanic("Unknown eviction policy in evictionPoolPopulate()");
        }

        /* Insert the element inside the pool.
         * First, find the first empty bucket or the first populated
         * bucket that has an idle time smaller than our idle time. */
        /* 取样的数据，计算好淘汰的idle后，放入淘汰池中 */
        k = 0;
        while (k < EVPOOL_SIZE &&
               pool[k].key &&
               pool[k].idle < idle) k++; // 自旋，找到淘汰池中比当前key的idle小的最后一个下标
        // k=0说明上面循环没进，也就是淘汰池中的所有数据都比当前数据的idle大，并且淘汰池的最后一个不为空，说明淘汰池也是满的，所以优先淘汰淘汰池中的数据
        if (k == 0 && pool[EVPOOL_SIZE-1].key != NULL) {
            /* Can't insert if the element is < the worst element we have
             * and there are no empty buckets. */
            continue;
        } else if (k < EVPOOL_SIZE && pool[k].key == NULL) { // 插入到桶后面
            /* Inserting into empty position. No setup needed before insert. */
        } else { // 插入到中间，会进行淘汰池的数据移动
            /* Inserting in the middle. Now k points to the first element
             * greater than the element to insert.  */
            if (pool[EVPOOL_SIZE-1].key == NULL) {
                /* Free space on the right? Insert at k shifting
                 * all the elements from k to end to the right. */

                /* Save SDS before overwriting. */
                sds cached = pool[EVPOOL_SIZE-1].cached;
                memmove(pool+k+1,pool+k,
                    sizeof(pool[0])*(EVPOOL_SIZE-k-1));
                // 假如当前数据比淘汰池的有些数据大，那么淘汰最小的
                pool[k].cached = cached;
            } else {
                /* No free space on right? Insert at k-1 */
                k--;
                /* Shift all elements on the left of k (included) to the
                 * left, so we discard the element with smaller idle time. */
                sds cached = pool[0].cached; /* Save SDS before overwriting. */
                if (pool[0].key != pool[0].cached) sdsfree(pool[0].key);
                memmove(pool,pool+1,sizeof(pool[0])*k);
                pool[k].cached = cached;
            }
        }

        /* Try to reuse the cached SDS string allocated in the pool entry,
         * because allocating and deallocating this object is costly
         * (according to the profiler, not my fantasy. Remember:
         * premature optimization bla bla bla. */
        /* 将当前的放入淘汰池 */
        int klen = sdslen(key);
        if (klen > EVPOOL_CACHED_SDS_SIZE) {
            pool[k].key = sdsdup(key);
        } else {
            memcpy(pool[k].cached,key,klen+1);
            sdssetlen(pool[k].cached,klen);
            pool[k].key = pool[k].cached;
        }
        pool[k].idle = idle;
        pool[k].dbid = dbid;
    }
}

简要看了一遍源码，我们对redis数据的淘汰机制有了一定的理解，并且知道淘汰算法有8种，所以下面主要介绍一下Redis中比较重要的LRU算法和LFU算法

Redis中的LRU算法

LRU，Least Recently Used翻译过来就是最久未使用，LRU算法根据使用时间淘汰数据，越久没使用的数据越容易淘汰。

• 实现原理

1. 首先，LRU算法是根据这个对象的操作访问时间来进行淘汰的，所以我们就需要知道这个对象最后的访问时间。
2. 知道了对象的最后访问时间后，我们就需要跟当前的系统时间进行对比，计算出这个对象已经多久没访问

• 源码验证

在Redis源码中，有一个redisObject对象，这个对象就是我们redis中所有数据结构的对外对象，它里面有个字段叫做lru

redisObject对象 (server.h文件)

typedef struct redisObject {
    unsigned type:4;
    unsigned encoding:4;
    unsigned lru:LRU_BITS; /* LRU time (relative to global lru_clock) or
                            * LFU data (least significant 8 bits frequency
                            * and most significant 16 bits access time). */
    int refcount;
    void *ptr;
} robj;

看注释，大概也能猜出来，redis去实现lru淘汰算法跟这个lru对象有关，这个字段大小为24bit，记录的是对象操作访问时候的秒单位的后24位(bit)，然后怎么获取秒单位的后24位？看一下例子：

long currentTimeMillis = System.currentTimeMillis();
System.out.println(currentTimeMillis/1000); // 获取当前秒
System.out.println(currentTimeMillis/1000 & ((1<<24)-1));// 获取秒的后24位

控制台打印一下，得到两个10进制参数

用二进制转换平台转换一下，1715915460二进制1100110010001101100101011000100

4639428二进制10001101100101011000100

两个参数对比一下，确实是拿到了最后24位

currentTimeMillis/1000 & ((1<<24)-1)为什么能获取到当前时间（二进制）的最后24位？还是画图看看，一个数和24个1进行二进制的与运算，就是获取最后24位数，如图所示

然后怎么获取24个1？细心的读者可能已经知道了，没错，就是(1<<24)-1，1左移24位再减1，如图所示：

二进制不熟悉，可以参考二进制运算

生活中的例子：

场景一：数据在5月份被访问，现在是8月份，我们可以通过8-3=5，得到这个对象3个月没访问

场景二：数据在5月份被访问，现在是3月份，我们可以通过：3+12-5得到这个对象10个月没访问

同理：

如果redisObject.lru < lruclock，直接通过lruclock-redisObject.lru得到这个对象多久没访问。

如果redisObject.lru > lruclock，直接通过lruclock+(24bit的最大值-redisObject.lru)

通过redis源码验证一下，发现源码的思路和我们上面所说是差不多的，查看estimateObjectIdleTime方法(evict.c)

/* Given an object returns the min number of milliseconds the object was never
 * requested, using an approximated LRU algorithm. */
unsigned long long estimateObjectIdleTime(robj *o) {
    // 获取秒单位时间的最后24位
    unsigned long long lruclock = LRU_CLOCK();
    // 因为只有24位，所有最大值为2的24次方-1
    // 超过最大值从0开始，所以需要判断lruclock(当前系统时间)跟缓存对象的lru字段的大小
    if (lruclock >= o->lru) {
        // 如果lruclock>=robj.lru，返回lruclock->lru,再转换单位
        return (lruclock - o->lru) * LRU_CLOCK_RESOLUTION;
    } else {
        // 否则，lruclock+(LRU_CLOCK_MAX - o->lru)，得到的对象的值越小，返回的值越大，越大越容易被淘汰
        return (lruclock + (LRU_CLOCK_MAX - o->lru)) *
                    LRU_CLOCK_RESOLUTION;
    }
}

Redis中的LFU算法

LFU，英文Least Frequently Used，翻译成中文就是最不常用的优先淘汰。不常用，它的衡量标准就是次数，次数越少的越容易淘汰。

• LFU的时效性问题

LFU算法有个问题需要去考虑，就是这个时效性问题，什么是时效性问题？就是去统计这个次数的时候，不能仅仅只考虑数量，而不考虑时间

举个例子，假如去年有一个新闻，很火，假如点击量是3000w，那么今年再有一个新闻出来，刚出来，点击量是1000w，本来我们应该让今年这个新闻显示出来的，去年的新闻虽然太火，但是也是去年的，我们推荐系统肯定不希望这个新闻继续上热搜的，所以推荐系统就需要考虑到数量同时兼顾这个时间问题

所以，如果根据LFU来做的话，仅根据使用次数来淘汰数据，很容易淘汰今年的新闻，所以容易导致新的数据进不去，旧的数据出不来，不过Redis里的LFU算法肯定是有考虑到这个问题的，具体是怎么实现的？

• 源码分析

来看redisObject的结构体，在server.h代码里，看里面注释，大概也知道在LFU算法的时候，里面这个lru，它前面16位代表的是时间，后8位代表的是一个数值，frequenct频率，应该就是代表这个对象的访问次数，我们先给它叫做counter

typedef struct redisObject {
    unsigned type:4;
    unsigned encoding:4;
    unsigned lru:LRU_BITS; /* LRU time (relative to global lru_clock) or
                            * LFU data (least significant 8 bits frequency
                            * and most significant 16 bits access time). */
    int refcount;
    void *ptr;
} robj;

前16bits代表时间，有啥用？跟时间相关，可以猜想应该和时效性有关。大胆猜测，这个时间是不是去记录对象多久没访问，如果多久没访问，就去减少对应的次数

找到Redis源码里的evict.c的LFUDecrAndReturn函数：

/* If the object decrement time is reached decrement the LFU counter but
 * do not update LFU fields of the object, we update the access time
 * and counter in an explicit way when the object is really accessed.
 * And we will times halve the counter according to the times of
 * elapsed time than server.lfu_decay_time.
 * Return the object frequency counter.
 *
 * This function is used in order to scan the dataset for the best object
 * to fit: as we check for the candidate, we incrementally decrement the
 * counter of the scanned objects if needed. */
unsigned long LFUDecrAndReturn(robj *o) {
    // lru字段右移8位，得到前面16位的时间
    unsigned long ldt = o->lru >> 8;
    // lru字段与255进行&运算，255代表8位的最大值，也就是二进制的8个1，得到8位counter值
    unsigned long counter = o->lru & 255;
    // 如果配置了lfu_decay_time，用LFUTimeElapsed(ldt)除以配置的值，总的没访问的分钟时间除以配置值，得到每分钟没访问，需要减少多少访问次数
    unsigned long num_periods = server.lfu_decay_time ? LFUTimeElapsed(ldt) / server.lfu_decay_time : 0;
    if (num_periods)
        // 不能减少为负数
        counter = (num_periods > counter) ? 0 : counter - num_periods;
    return counter;
}

redis配置

lfu-decay-time 1 // 多少分钟没操作访问就减1次

而对应8bit的次数，最大值是255，可以看下redis源码LFULongIncr函数，在evict.c

/* 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) {
    // 如果等于255，直接返回255,8位的最大值
    if (counter == 255) return 255;
    // 得到随机数，0到1之间
    double r = (double)rand()/RAND_MAX;
    // LFU_INIT_VAL表示基数值，默认为5，在server.h配置
    double baseval = counter - LFU_INIT_VAL;
    // 如果当前counter小于基数，那么p=1，r肯定小于p，所以counter肯定加
    if (baseval < 0) baseval = 0;
    // 不然，按照几率来校验counter是否加，跟baseval和lfu_log_factor这两个参数相关，因为都是在分母，所以两个值越大，p越小，也就是counter++的概率越小
    double p = 1.0/(baseval*server.lfu_log_factor+1);
    if (r < p) counter++;// p越小，counter++的几率就越小，反之亦然
    return counter;
}

所以，LFU的实现逻辑，可以总结一下：

1. 如果达到255最大值，counter就不加，因为达到255的几率不是很高，可以支撑很大的数据量
2. counter是随机添加，添加的概率和已有的counter值和配置的lfu-log-factor两个参数相关，已有的counter值越大，添加的几率越小，配置的lfu-log-factor值越大，添加的几率也越小

在redis官网找到如图的压测数据图，里面facror就是配置的lfu_log_factor，可以看到配置的值越大，需要达到255的最大值就需要更多的hits