DISTRIBUTED DATA PROTECTION: New Approaches to an Old Challenge

by Greg Schulz
April 1, 2005

 

Data Mirroring and Replication

There are several approaches to data duplication, including synchronous (real-time) and asynchronous (time   delayed), implemented in hardware and software. Commonly referred to as remote mirroring or replication, these forms of data duplication involve maintaining at least one copy of data locally or remotely. Many storage subsystems can mirror and replicate data externally between two or more storage arrays, usually from the same manufacturer (homogenous). Data can be moved and accessed over distance using technology, including storage over IP encompassing Internet FCP (iFCP), Fibre Channel over IP (FCIP), and iSCSI; storage over Synchronous Optical NETwork (SONET), Synchronous Digital Hierarchy (SDH), and Course Wave Division Multiplexing (CWDM) or Dense Wave Division Multiplexing (DWDM) using fiber optic wavelength, or dedicated fiber optic services.

Remember that if your data protection strategy is to simply rely on mirrored data, should the data be deleted or corrupted, it’s deleted and corrupted on both the primary and secondary copies. A solution is to take a mirrored data volume offline and perform backup for data protection. A good combination is to use synchronous for relatively short distances of about 100km to a secondary site and asynchronous data replication to a third site that could be 1,000km or further away. When configuring remote storage interfaces, use redundant paths utilizing diverse networks. These could be with the same or different carriers. Remember, the two paths take divergent routes.  

 

Data Retention and Archiving

Archiving is moving infrequently accessed and soon-to-be-deleted data to another storage medium. Backup and mirroring are mechanisms to protect against loss of data and loss of access to data. An advantage of data archiving is that, by removing infrequently accessed data, less data needs to be backed up, speeding backup and replication functions. Data retention is concerned with the long-term retention of data on some form of storage medium. Data may be written in Write Once, Read Many (WORM) or standard read/write mode using magnetic disk, tape, or optical storage mediums. Data may also be archived to near-line storage such as NAS file servers or iSCSI and Fibre Channel Serial ATA (SATA)-based storage subsystems.

Storage and Data Protection Over Distance

Network bandwidth is becoming more readily available at faster rates and lower prices. Unfortunately, the amount of data that needs to be moved, the distances involved, and the number of locations continues to rise. Consequently, networking costs remain the same or potentially increase to support the larger amount of data that needs to be moved. Bandwidth options range from fiber optic, T3, ISDN, xSDL, cable, WiFi (fixed and roaming) and satellite, among others.