Using timestamp to implement database to process the time difference problem of the same record __ database

At work, we often process a row of records to find that there are multiple threads at the same time (SQL Server), although the system takes the appropriate locking policy to prevent contention deadlock, but for a frequent processing of records, due to the system's multithreaded mechanism, So when multiple transactions process the same record, because the transaction is of different size and time is sequential, this may result in dirty reading of the database, and it is found that using timestamp is a better solution by studying multiple methods, that is, adding a column of fields to the most frequently used table, the type is timestamp, After the addition of the system will automatically generate the corresponding unique value, if the data records have any changes, timestamp value will be adjusted accordingly, using this method, we at the beginning of the transaction process, the first record of the old timestamp value, such as real changes in the data, check the timestamp is the same, If the same, the data is not dirty to continue processing, if the difference is that the processing of other threads before the modification, this time can be discarded to ensure that the data security consistent.