When you encounter a problem analysis record at work, you should directly throw the problem. Existing layer-3 business A, B, and C raw data layer-A stores the basic data object map =>... the size is no less than 1000. A data object (A1) is randomly thrown at the underlying data layer B. After capturing the object A1 thrown at the data layer A, it is processed by a series of DB queries and in-depth algorithm computing, generate the result B1 corresponding to A1. Layer B throws data to layer C of the business, captures layer B data, and processes the business. The problem arises. If layer a throws the object A1. Layer B consumes a lot of resources to process each time to produce B1. To solve this problem, I think of three solutions, analysis and recording. The secret is to choose based on the advantages and disadvantages according to the actual situation (third type is not recommended by the monks )~ 1. A-> B-> C each time following the normal process. Advantages: no additional storage overhead is required, and the overall process is clear and simple. Disadvantages of timely data update: computing, high query overhead, result generation speed, dependent on network and computing difficulty 2. maintain a map {a1_key => B1} in layer B. Advantages: clear structure, computation, low query overhead, and fast result generation. No data pollution. Disadvantage: One more map needs to be maintained, and the latest dB data is not updated in a timely manner. 3. Mount the results in layer B to object A1. Advantages: computation, low query overhead, and fast output. Access is extremely convenient. Disadvantage: Data pollution on Layer A is not easy to maintain.
Advantages and disadvantages of Data Object Storage