USING REPLICATED DATA TO REDUCE BACKUP COST IN DISTRIBUTED DATABASES

To increase fault tolerance in distributed database, it is better to add a backup server for each primary server in the system. It is clear that the primary server and backup server need to be connected to each other. To connect these computers to each other when they are in a long distance, it is necessary to use a lease line which needs to be charged as data is transferred. As more packets are transferred between primary and backup server, more money need to be paid for charging this line. So if number of transferred packets between these computers reduces, the company can economize in its expenditures. On the other side, when number of updating information from the primary server to backup server reduces, the number of transaction which should be performed in the backup server reduces. To achieve this goal, we introduce a new method which reduces the number of transferred packet between primary and backup server. In this method, the replicated data of primary server is used to backup mechanism. In our method the primary server sends transactions in data which are not replicated in other computers. So the transactions on the replicated data are not transferred to backup server, and as a result the numbers of transferred packet get reduce