About replication

Replication allows you to send data from a primary DS-System to other DS-Systems in a DS-System replication group. You can add as many DS-Systems (Windows or Linux) as you want to a DS-System replication group, and each DS-System will have a copy of the shared DS-Client data.

You need at least two DS-Systems to perform replication. A FullFeatured DS-System can replicate DS-Client data one way to a replication DS-System. Two FullFeatured DS-Systems can replicate shared DS-Client data to each other. Only backup data replicates from one DS-System to another. Items that are not replicated include logs and DS-Client configuration settings. By default, DS-Systems in a replication group communicate with one another on port 4409.

Once a DS-System group is configured and is replicating, the corrective capabilities are multi-directional. This means that any data loss on one DS-System can be corrected with the data from another. When a DS-Client connects to a replication DS-System, it can continue to restore data as if it were connecting to the primary DS-System. The only condition is the replication DS-System must be up-to-date with the production DS-System data. For more information, see “Clearing the replication flag”.

To maximize the replication speed, DS-Systems in a replication group send only deltas after the first master generation of a file is replicated (replication is “incremental forever”). For example, a backup file with 100 generations even if it has 1 master every 10 generations on the primary DS-System, would only replicate the first master file and then send 99 delta generations. Since each DS-System is responsible for managing its own backup data, the task of optimizing what is stored is left to each replication DS-System. Because replication sends one master and only deltas for a backed up file, each DS-System that receives replicated data may encounter long delta chains (e.g. 20 or more deltas in a row after a master). Since this may create a performance issue in the event you need to restore, the DS-System has a built-in optimization capability that it runs on schedule. By default, the DS-System schedules and runs a delta chain optimization task once a day. For more information, see “Configuring the delta chain settings”.

Since some backup files can be very large, replication has the capability to resume at the block-level from failed transmissions of replicated data. Block-level resume is performed for files that are over 10 MB in (stored) size on the DS-System online storage. Block-Level Resume allows the replication to continue from the last successful file block transmitted to the replication DS-System. The benefit increases with the size of the file being replicated. Without this feature, replication of a large file that is interrupted because of a slow / poor connection would have to restart from the beginning of the file.

The protected size always remains the same for each replicated DS-Client's data. However, replication tries to optimize on transmission time and storage, which can result in the stored size being significantly different between the primary and replication DS-Systems. For this reason, an invoice generated in DS-Billing for a replication DS-System might show different storage numbers than an invoice generated in DS-Billing for the primary DS-System. The following are some possible reasons:

• Library files are replicated from a primary DS-System to a replication DS-System as libraries and their information is inserted in the DS-System database of the replication DS-System so that future files with the same content will link to them. The invoices in DS-Billing for the primary DS-System are calculated by adding a library amount once for each DS-Client that links to that library. If only some of the DS-Clients are replicated from the primary DS-System, those differences will show up in the invoiced amount.

• Incremental forever replication is performed from the primary DS-System. Only the first generation of a file that qualifies for master/delta processing is replicated as a master generation (including all data blocks). Incremental generations are replicated as delta generations. In certain scenarios, the primary DS-System might store more master generations for the same file than the replication DS-System. Those differences will show up in the invoiced amount (stored size). All new DS-System installations will ignore recycled delta generations when calculating 'stored size'. You can configure this behavior separately, for each DS-System in the replication group using the ExcludeRecycleDelta parameter. For more information, see “Configuring the advanced settings”.

For customers who need to recover a large quantity of data, the disc/tape feature may not provide a fast enough solution. For these cases, a replication DS-System allows you to use that replica of the production DS-System to be physically sent to the remote location. It can be used as follows: