Counter pattern instance of Redis's usage pattern

Redis is the most popular Fried chicken in NoSQL. It is like a Swiss army knife. It is small, sharp and practical. I plan to write a series of articles on Redis usage patterns, which will provide an in-depth summary of common Redis usage patterns for your reference.

common summary counter

Summary counting is a common function of the system, such as the website usually needs to count the number of registered users, the total number of website visits and so on. Summary counters can be implemented using the basic data types provided by Redis, and increment operations can be implemented through the incr command.

For example, the number of registered users, the basic operation command is as follows:

 #  Get the number of registered users 

  get total_users

  #  Increase of registered users 1 position 

  incr total_users

counter by time summary

Counting is usually done by time. For example, the number of registered users needs to be counted by day.

Or the registered user count example, the basic operation command is as follows:

#  Assume that the operating  2014-07-06  data 

  #  Get the number of registered users 

  get total_users:2014-07-06

  # 2014-07-06  Increase of registered users 1 position 

  incr total_users:2014-07-06

  #  Set up the  48  Hour expiration time  172800 = 48 * 60 * 60

  expire total_users:2014-07-06 172800

Setting a 48-hour expiration time for the counter is to save space for the counter. After all, redis is an in-memory database, so you can perform one task before expiration to store the counter in a relational database.

speed controls

Speed control is also one of the common counting purposes of Redis 1. For example, there is an API service, and you want to control the number of requests per second per IP not more than 10 times. You can use IP and time seconds as key to set a counter to achieve control.

 #  Maximum number of requests per second 

  MAX_REQUESTS_PER_SECOND = 10
  #  check  ip  Request limits 

  # @param ip

  # @raise  Over the limit, throw  RuntimeError  abnormal 
  def check_request_limitation_for_ip(ip)

    time_tick = Time.now.to_i

    key = "#{ip}:#{time_tick}"

    num = $redis.get(key).to_i

    if num > MAX_REQUEST_PER_SECOND

      raise 'too many requests'

    else

      $redis.incr(key)

      $redis.expire(key, 10)

    end

  end

maintains a large number of counters, , using the Hash data type

Sometimes need to maintain a large number of counters, such as every BBS theme 1 check number, each user to access the page number 1, for example, because the BBS topics and user base can be large, directly based on BBS theme or user ID generate counter, occupy or considerable Redis resources, then can use Hash data type compress the resources needed.

For example, the view count for the corresponding forum topic can be set by mode

  key: topic:<topic_id>:views

  value: view count (integer)

Convert to mode:

 key: topic:views

  value: hash

    hash key: <topic_id>

    hash value: view count (integer)

Conclusion: with Redis counter, various counting functions can be realized simply and efficiently.