- Home
- Products & Solutions
- Storage
- Solutions
- Remote Backup & Replication
What is Remote Backup & Replication?
Equipment and applications can be replaced, yet digital assets can be irreplaceable, so the minimum disaster recovery (DR) protection that companies require, is to make backup data safe from loss or damage. When all copies of data are stored on a single hardware system, they are vulnerable to the same fire, flood, or other risks.
For most users, removable media that can be removed off-site provides site loss protection. One of the main reasons that disk backup isn’t used more, is that backup sets are so large and network bandwidth too limited to allow the replication of backup data over networks.
However the introduction of data deduplication is making the remote replication of backup practical. Deduplication ensures that only data blocks that have changed since the last backup are replicated. For instance if only 1% of a 500GB backup has changed since the previous backup on the source system, then only 5 GB of data has to be moved to the target system to keep the two systems synchronised. This data can be moved in the background over several hours, meaning you can use a standard WAN (Wide Area Network) to replicate backup sets.
To reduce the amount of time the first replication takes, devices can be co-located initially and the data moved over a faster network, or backup data can be put on tape at the source site then transported and copied locally onto the target system. After that first synchronisation, the replication only needs to move the new data blocks.
For disaster recovery, this means you can have an offsite replication of all your backup data every day, which provides better protection. Plus, disk based deduplication and replication reduces the amount of removable media required lowering costs and risks. Most IT departments combine the technologies, using tape for the longer-term retention of data backup, as it provides the right economics and portability.
Replication Approaches – Synchronous and Asynchronous
Synchronous replication, also referred to as mirroring, continuously maintains two primary datasets in the same state by transferring blocks between two storage systems at each I/O (input/output) cycle. Synchronous replication is normally designed to provide a very rapid automatic switch over to the replica if the primary dataset is compromised, and it usually involves two separate storage systems, often in different locations. Because synchronous replication systems delay I/O-complete status signals to the host until both the local and remote writes are complete, they require high-speed links, although performance is always reduced and they are more complex to manage. Therefore synchronous replication is typically reserved for very high value primary data used in transaction-oriented applications that must remain continuously available.
Asynchronous replication maintains a second data set dynamically, as a duplicate of the primary set, but it is allowed to lag behind the primary by some period of time. The delay may only be one or two I/O cycles (although it may be longer), so the mirror is a near replica. Asynchronous mirroring requires less bandwidth and normally minimises negative impact on operation of the primary data, although if the mirrored image falls too far behind the primary, the primary system may have to periodically suspend writes to allow the mirror to catch up.
Asynchronous replication can also be applied to non-dynamic, point-in-time images, including backup images, to provide site loss and disaster recovery protection. It is much less complex to implement than mirroring techniques, can provide protection from other classes of faults, reduce an organisation’s use of removable media and it has less impact on primary applications.
For more information:
What else would you like to see on this page?
 |
 |
 |
 |
|
|
|||
