1. Moongo db will try to compare all the indexes and hot data into memory to get a better query speed, while MongoDB writes, it is written in memory, and then periodically synchronized to disk, so that can achieve the high performance of sequential writing, So MongoDB is a memory-eating thing; it is best to deploy MongoDB on a separate server; You can limit the size of the memory with Ulimit; 2. Because MongoDB takes full advantage of the memory features. If your index and the size of the hot data is more than the size of the memory, then the performance of the reading is not very good, because of the good frequency of memory paging (there may be a change of index may also have the possibility of exchanging data) operation; 3. Mongostat can be used to view the indicators of MONGO; 4. The amount of data to a machine can not withstand, causing performance problems when the solution: (1) According to their own business regularly delete data, (2) do Shard Sharding, (3) Improve memory, (4) hard disk to do SSD, the actual production environment, replaced by SSD better;
MongoDB several issues to be aware of
