One: Memcached cache avalanche Phenomenon
(1) Cause: Usually by a node failure, resulting in other nodes cache hit rate drop, the missing data in the cache query, a short period of time caused by the database server crashes. Restart the DB, the short-term is also crushed, but the cached data increased some, DB repeated multiple start, multiple caches can be established, the DB can be stable operation.
Or because the cache is periodically invalidated, such as every 6 hours, every 6 hours, there will be a request ' peak ' severity that will even crash the db.
(2) Assuming 1200 concurrency per second, the Memcached cycle expires every 6 hours, due to the excessive amount of requests memcache is full, resulting in BD collapse.
Solution:
1: Set the cache to a random 3-9-hour life cycle, so that different failures, to assign work to the various points up.
2: The wedding car time to increase, to the early morning 4-5 [the minimum number of times] to update the cache.
One-Memcached cache avalanche phenomenon