Server performance RAM: 4 GB/1067 MHz/DDR 3cpu: 2.53 GHz Intel Core 2 Duo insertion: 0.1 million 25 seconds after the same record is put, the TT file is more than 400 mb, however, MongoDB is 2 GB in size. Therefore, it is assumed that this is the completion of the MongoDB space pre-allocation to obtain records through the primary key: Every 0.1 million times 33 seconds through the Primary Key Update: Every 0.1 million times 25 seconds after 5 tests, we can obtain MongoDB performance indicators: 5 concurrency, the average throughput of 1.1 million records inserted: About 6600 records/second, 10 concurrency, and the average throughput of 2.2 million records inserted: about 6300 concurrent records per second, the average throughput of 2.2 million records inserted: about 6300 concurrent records per second, 50 concurrent records inserted, and the average throughput of 5.5 million records inserted: about 4500 concurrent records per second, with an average throughput of 100 records inserted: about 5.5 million records per second MongoDB is actually a memory database, which first saves data to the memory, then write the data to the disk. MongoDB can use the object class in the document creation, or document. 1) if it is an object class, you can only use the LINQ method during query. Or document objdocumentquery = new document {"orderid", op. greaterthan <int> (2) }}; icursor <test1> it = collection. find (objdocumentquery); that is to say, OP must be used for dynamic queries. 2) If it is document, you can use linq, document (10), you can change to 1000 mongod.exe -- dbpath F: \ MongoDB \ mongodb-win32-i386-1.8.2 \ data -- rest -- maxconns by command to find the status document. sendcommand (new document (). append ("serverstatus", 1); connection string Server = 127.0.0.1; connecttimeout = 5000; minimumpoolsize = 8; maximumpoolsize = 256; pooled = true