The implementation of similar second kill by Redis+lua

The implementation of similar second kill by Redis+lua

Due to the need of the project, multi-thread is needed to obtain and modify the inventory of the database. Considering the low efficiency of locking the database, redis+lua is used to implement it.

Redis's single-threaded operation feature executes lua scripts and ensures atomicity through lua scripts. If you make changes through simple redis instructions, there will be the problem of multithreaded concurrent updates between reading and writing.

1. First define the redis data structure

GoodId: {"total": 100, "released": 0;} where goodId is the commodity id number, according to which you can query the relevant data structure information. Total is the total number, released is the quantity issued, and available number is total-released.

two。 Write lua scripts local n = tonumber (ARGV [1]) if not n or n = 0 thenreturn 0endlocal vals = redis.call ("HMGET", KEYS [1], "total", "released"); local total = tonumber (vals [1]) local blocked = tonumber (vals [2]) if not total or not blocked thenreturn 0endif blocked + n database

In view of the update from redis to databases, I have thought about it for a long time, but I haven't found a better solution. I first adopt the asynchronous update of scheduled tasks. As for the question of whether data is lost, if redis dies, redis will recover the data after restart, and the data in the database can be kept consistent with the next scheduled task. The disadvantage is that if redis fails, flash sale activity will fail.

As for how to drive redis to database updates, list two folks:

Redis keeps a list of related hash keys and reads updates by reading the list table and reading updates by streaming the tables in databases.

You are welcome to ask questions. Thank you for reading.