Software-based point-in-time copies, or snapshots, enable a number of benefits, including replication between heterogeneous storage systems.
By Karen Dutch
There is more to replication than mirroring for disaster recovery. Although replicating data can reduce recovery time, making point-in-time (PIT) copies, or snapshots, of data can also be invaluable to IT organizations. The convergence of affordable disk capacity, bandwidth, and replication software has resulted in a variety of products that can increase an organization's operational efficiency and agility.
Data replication refers to the copying or mirroring of data from one system to another system. In the past, replication was done only at the hardware level inside the disk array. While this approach allowed users to make copies of critical data, it limited them to replicating data between storage systems from the same vendors and, in some cases, between systems of the same make and model.
Today, replication can also be done at the software or host level and, on a limited basis, in the network or storage area network (SAN) fabric. Software-based replication in many cases allows organizations to replicate between unlike storage systems, and because no pre-requisite hardware or software is needed, it allows administrators to quickly distribute the benefits of replication to a variety of systems within the organization. In this way, organizations can tap into low-cost servers, storage, and bandwidth.
Software-based replication can make it extremely cost-effective to replicate data anywhere over TCP/IP. The question is: What do you do now that you can make fast, low-cost copies of your data and send them anywhere?
The value-add of replication lies with the ability to make PIT copies or snapshots of data. Snapshots are nearly instantaneous images of stored data taken at a specific moment in time. Snapshots can be used to enhance other backup processes, to help consolidate resources, to aid in data migration, or to improve application testing or analysis, and their value is often directly tied to cost-saving or revenue-producing activities.
Backup consolidation—As the amount of stored data increases exponentially, backup becomes a complex—and costly—undertaking. Using a many-to-one replication configuration, companies can consolidate PIT copies of data onto a single backup server. This also improves overall disaster-recovery capabilities since PIT snapshots are stored on disk, not tape. Doing so simplifies the backup architecture and reduces the number of dedicated tape devices and associated backup licenses. The combined effect is lower backup costs.
Windowless backup—Today's 24x7 environment makes it difficult for organizations to halt transactions to run backups—even late at night. Similarly, online database backups have created their own difficulties (e.g., poor application performance during backups. Organizations can avoid interrupting production operations by making snapshots of data and then replicating data to tape from the PIT copy.
Resource consolidation—Organizations are looking to centralize and consolidate server, storage, and data-center consolidation to improve overall IT efficiency. PIT snapshots play an essential role in allowing organizations to quickly and easily move associated data and files to the new consolidated environment.
Data migration—Similar to resource consolidation, snapshot replication plays a key role in helping organizations move data to a new, consolidated storage environment. This kind of migration is necessary almost every time storage or servers are upgraded. By using PIT snapshots, organizations can quickly and non-disruptively migrate data from one array to another regardless of the array's make and model.
Application development and testing—Application and system developers can use PIT snapshots to test new systems using real data and without impacting the production system. By working with near-exact copies of current data, developers get a real sense of how well an application or system will run in actual production. This helps reduce application development time and improve application robustness.
Data analysis/reporting—Similar to application development and testing, PIT snapshots can be used to feed nearly current production data to analytical systems for data mining, etc., purposes. PIT snapshots allow analysts to run complex queries on real data without imposing any overhead on the production system.
You can create PIT snapshots of data for a wide variety of purposes. Snapshots can be used to create a copy of data so that the data is closer to where it will be used. An organization may create and save a sequence of snapshots over time so that it can restore files or data from various points in time in the past.
The power of software-based replication lies with its flexibility. There is no limit to how many copies of data you can make. And the ROI increases each time you access/use one of the snapshots for any of the purposes detailed above. The ROI of such uses can justify the purchase of replication software even without some sort of unplanned DR event.
Karen Dutch is vice president of product management at Softek (www.softek.com) in Sunnyvale, CA.
For more information about replication,
see Parts I and II of this three-part series:
Part I, "Data replication: Host, network, or array,"
Nov. 2003, p. 40
Part II, "Leveraging replication to improve disaster recovery,"
Dec. 2003, p. 32