In an e-commerce or service platform, the(computing) cacheuse is key to improving system performance and responsiveness. However, thecache passthroughis a common performance bottleneck problem, especially when querying non-existent data, the system will access the database directly, which not only affects the performance, but also may cause the database to be overburdened. In order to solve this problem effectively, we propose a combination of theBloom Filter、Null Cache cap (a poem)distributed lock The cache-penetration protection scheme. The following is the workflow of the program.
workflow
1. User request for coupon template information
The user first initiates a request for coupon template information. The request includes a coupon template ID, based on which the system needs to return the corresponding coupon information.
2. Cache queries: Redis caching
The system starts with theRedis Cache Redis is an efficient caching system that can often greatly improve query speed. If the corresponding template information exists in the cache, the system returns it directly to the user and the query process ends.
3. Cache misses: the use of Bloom filters
If the corresponding coupon template information is not found in the Redis cache, the system further passes theBloom Filter Check if this template ID is valid. Bloom filters are a space-efficient data structure used to quickly determine whether an element is in a collection.
- If this template ID is not present in the Bloom filterIf the coupon template ID is not valid or has been invalidated, the system will return it directly to the user."Failed: invalid coupon template ID"。
- If this template ID exists in the Bloom filter, indicating that the coupon template ID may be valid and the system will continue to query the database.
4. Null caching: preventing duplicate queries
If the Bloom filter determines that the template ID is valid, the system continues to check the Redis cache for null caches. Null caching means that for some queries, the database returns "null" results (e.g., a coupon template ID that does not exist in the database), and these null results are cached for a certain period of time in order to avoid querying the database repeatedly.
- If a null value exists in the Redis cacheThe system will directly return the"Failed: invalid coupon template ID", avoiding repetitive database queries.
- If there are no null values in the Redis cacheThe system continues to perform database query operations.
5. Distributed locks: ensuring data consistency
To prevent multiple requests from querying the database at the same time, resulting in excessive database pressure, or multiple threads performing the same query operation at the same time, the system uses thedistributed lock to ensure that only one request will access the database at a time to query data.
-
If distributed locks are available, the system acquires the lock and performs the following steps:
- Query the database for coupon template information.
- If the database returns data, the system caches the data into Redis, reducing access to the database for subsequent requests.
- If the database returns empty data, the system caches the empty results in Redis and sets a short time expiration to prevent repeated queries within a short period of time.
- Finally release the distributed lock.
-
If distributed locks are not available, indicating that other requests are performing the same database query operation, the system waits for the lock to be released or returns an error message.
6. Return results: cached or database data
- If there is data in the Redis cache, the system returns the cached data directly to the user.
- If there is no data in the cache and the query succeedsThe system returns data from the database to the user and caches that data to improve the efficiency of subsequent queries.
- If the query fails(e.g. invalid template ID or no data in the database), the system returns an error message.
flow chart
code implementation
public CouponTemplateQueryRespDTO getCouponTemplate(CouponTemplateQueryReqDTO requestParam) {
// Query if the coupon template information exists in the Redis cache.
String cacheKey = (RedisConstants.COUPON_TEMPLATE_KEY, ());
Map<Object, Object> cacheMap = ().entries(cacheKey);
// If the cache exists, return it, otherwise query the database through Bloom filters, null caches, and distributed locks.
if ((cacheMap)) {
// Determine if the Bloom filter exists for the specified template ID, if not, return an error
if (! (())) {
throw new ClientException("Coupon template does not exist");
}
// Query the Redis cache to see if a null value exists, and return it if it does.
String nullCacheKey = (RedisConstants.COUPON_TEMPLATE_NULL_KEY, ());
Boolean isNullCached = (nullCacheKey);
if (isNullCached) {
throw new ClientException("Coupon template does not exist");
}
// Get the distributed lock
RLock lock = ((RedisConstants.LOCK_COUPON_TEMPLATE_KEY, ())); ((RedisConstants.
();
try {
// Double-check the null cache
isNullCached = (nullCacheKey);
if (isNullCached) {
throw new ClientException("Coupon template does not exist");
}
// Use double-checked locks to avoid concurrent queries to the database
cacheMap = ().entries(cacheKey); if ((cacheMap) {
if ((cacheMap)) {
LambdaQueryWrapper<CouponTemplate> queryWrapper = ()
.eq(CouponTemplate::getShopId, (())))
.eq(CouponTemplate::getId, (())))
.eq(CouponTemplate::getStatus, ());
CouponTemplate couponTemplate = (queryWrapper);
// If the template does not exist or has expired, set the null cache and throw an exception
if (couponTemplate == null) {
().set(nullCacheKey, "", 30, ); }
throw new ClientException("Coupon template does not exist or has expired");
}
// Serialize the database record into the Redis cache.
CouponTemplateQueryRespDTO responseDTO = (couponTemplate, );
Map<String, Object> responseMap = (responseDTO, false, true);
Map<String, String> cacheData = ().stream()
.collect(((
::getKey, .
entry -> () ! = null ? ().toString() : ""
));
// Use a Lua script to store the data into Redis and set the expiration time.
String luaScript = "('HMSET', KEYS[1], unpack(ARGV, 1, #ARGV - 1)) " +
"('EXPIREAT', KEYS[1], ARGV[#ARGV]) ";
List<String> keys = (cacheKey).
List<String> args = new ArrayList<>(() * 2 + 1).
((key, value) -> {
(key);
(value);
});
// Set the expiration time for the coupon campaign
((().getTime() / 1000)); // Set the expiration time of the coupon campaign.
// Execute the Lua script
(
new DefaultRedisScript<>(luaScript, ),
keys, (); // Execute the Lua script.
()
);
cacheMap = ()
.stream()
.collect((::getKey, ::getValue));
}
} finally {
(); }
}
}
// Return the data retrieved from the cache
return (cacheMap, , false, ()); }
}