SAN-based data replication.
Data Replication Alternatives
Although a heterogeneous storage infrastructure undoubtedly increases management complexity, it also provides IT managers with important benefits. Maintaining multi-vendor relationships often leads to significant cost savings, owing to the leverage gained when negotiating new storage contracts. And heterogeneous environments increase a storage manager's flexibility, allowing end-user allocation requests to be satisfied using the most appropriate technology, regardless of vendor. Achieving cost-effective business continuity/disaster recovery (BC/DR) configurations in a multi-vendor environment depends on interoperability support from data replication solutions.
Utilizing the replication technology provided by a high-end array vendor delivers unparalleled failover functionality, and supports the most demanding application BC/DR needs. Array-based solutions leverage the controller of the storage array as an operating platform for replication functionality. The tight integration of hardware and software gives the storage vendor unprecedented control over the replication configuration, and allows for service-level guarantees that are difficult to match with alternative replication approaches.
High quality of service comes at a cost, however. Array-based replication invariably sets a precondition of like-to-like storage device configuration. This means two similarly configured high-end storage arrays must be deployed to support replication functionality, which increases costs and ties the organization to a single vendor storage solution.
Running replication functionality from the application server addresses some of the disadvantages of array-based solutions but introduces others. Server-based replication software functions with any storage hardware that can be mounted to the application platform, offering heterogeneous storage support. However, server-based solutions can be costly to license and manage, as software is required on each server and backup server. Operating from a general-purpose server platform and sharing resources with the business application also imposes performance penalties, making it appropriate for only low-end applications.
SAN-Based Replication: A New Alternative
Like many storage management applications, data replication functionality is finding its way into the network. SAN-based replication solutions bridge the gap between low-end server-based, and high-end array-based approaches, and support a variety of storage services, including local and remote synchronous mirroring, asynchronous mirroring over long distances, point-in-time snapshot replication and rollback capabilities, providing a variety of local and remote data replication alternatives.
SAN-based replication solutions support heterogeneous storage devices and heterogeneous server platforms, giving the widest possible range of configurations. By enabling any-to-any connectivity, SAN-based solutions deliver the flexibility storage managers need to implement a tiered-storage model. Support for heterogeneous connectivity effectively decouples the performance and availability features of high-end storage arrays, and enables business applications to be mapped to the most cost-effective storage solution, regardless of vendor.
The Advantages of Network-Based Intelligence
Storage infrastructure total cost of ownership (TCO) is a hot topic within IT. According to a recent survey of CIOs, cost reduction was a key goal for 2004, driving initiatives for tiered storage architectures and Information Lifecycle Management (ILM). Storage-related expenses for a typical large organization account for as much as 25% of IT spending. And the forces driving these costs show no signs of abating.
So, how does network-based intelligence help IT reduce costs? Network-based intelligence reduces capital and operating expenses on a number of fronts:
* Enables a tiered storage approach rather than a one-size-fits-all approach, mapping application needs to the most cost-effective storage
* Enables Data Lifecycle Management, moving data to less expensive media as it ages or as access rates change
* Enables consolidation, potentially reducing the amount of storage resources required
* Provides a single point of management for replication services, reducing or negating the need for multiple back up/restore tools and licenses
* Moves backup off the host and storage, making them more available for production, and effectively reducing backup windows to zero
* Restores are faster and 100% successful
Tiered Storage/Data Lifecycle Management
Tiered-storage takes the presence of heterogeneous storage resources as a given. It provides storage managers with the means to categorize and prioritize business applications by the level of service they demand, including Recovery Time Objectives and Recovery Point Objectives for BC/DR planning, and then map those requirements to the appropriate class of storage using a tiered architecture. The result is efficient allocation of all resources and cost-effective BC/DR.
For instance, many data centers today deploy a two-tier data protection practice: Tier I storage is used for all production, backup copies, disaster recovery/fail-over copies as well as any copies required for development and test environments. Tape is typically used for archive copies. In all, a single application may require five or more copies of Tier I storage to adequately protect a single production copy. With homogeneous storage replication, these copies must all occupy the same type (and cost) of storage as the most used primary copy.
A network-based replication solution, on the other hand, enables the implementation of a multi-tier storage strategy. In this case, heterogeneous storage replication methods allow the use of storage that specifically meets the needs of the data in each stage of its lifecycle. For instance, production data continues to be written to and read from Tier I storage to provide the highest reliability and performance requirements of enterprise applications. Because the replication is now done in the network, the second copy of data does not need to reside on Tier I storage but can reside on more cost-effective Tier II storage. For the DR environment, Tier II storage provides a good fit, providing enough performance to support a fail-over and restore environment at roughly half the cost of Tier I storage. Tier III storage like ATA, which is a fraction of the cost of Tier I storage, can be a good solution for development and test environments. These environments frequently run multiple versions of data for analysis, making Tier I storage very costly. ATA storage can also be a good fit for backup, providing a gateway to tape archive and a standby restore platform. The result is a significantly reduced total cost of ownership for the storage infrastructure.
In addition to lower storage acquisition costs, tiered storage architectures also significantly reduce operational costs. Typically, annual maintenance is a percentage of the price of each storage frame. The lower the cost of the storage frame, the lower the annual maintenance, dramatically reducing ongoing, operational costs.
Once you have an environment that supports tiered storage, you can begin to deploy data lifecycle management solutions. These solutions allow the automatic migration of data to less expensive storage as it ages, based on established policies. By migrating older data off to archive in a timely manner, you can meet regulatory data retention requirements while managing less data and reducing overall management costs.
Network-based intelligence also enables consolidation and improved operational efficiencies. Rather than having a variety of backup environments--each with their own backup and restore tools and processes--you can consolidate the backup environment to the network, dramatically reducing the number of licenses required to be maintained. It also allows you to consolidate storage and use a shared tape library and media server across all environments rather than having dedicated storage for each environment. As an example, most companies have a mix of Windows and Unix environments, each with their own backup processes. Backup and restore tools are typically unique to each environment, so you end up with multiple backup server licenses for each local and remote backup server. Equipment such as tape libraries and media servers are also dedicated to applications, making efficient sharing impossible. By moving the backup process to the network, agents on individual hosts for backup and restore are consolidated to a single media server, reducing the number of licenses required and simplifying management.
Improved Backup and Recovery
Leveraging low-cost ATA disk as a gateway to archive and as a standby restore platform also improves backup and recovery. Archive to tape requires a stable copy of data to be available at a specific point in time. The intelligent backup capabilities of SAN-based replication appliances allow an application group to be snapped from the production storage to the backup storage and a specific version of the data presented to the media server for archive to tape. This has several benefits for backup and recovery:
Storage Cost: Utilizing ATA storage instead of Tier I storage to interface with the media server provides a huge savings in storage acquisition and operational costs--with ATA typically 1/10 the cost of Tier I storage.
Restore Platform: Most applications require split copies of the data to protect against logical errors. In addition, application data must be backed up daily and brought off site. Leveraging low-cost ATA disk as the restore platform, and saving the last 2 weeks of data on that restore platform, means that restores can now happen from disk instead of tape. Because the 14 days of data protect against logical errors, you can reduce the number of archive copies you put to tape and take off site, providing significant savings in both tape transport and media costs.
Software Costs: Agents on individual hosts for backup and restore are consolidated to a single media server, potentially reducing or negating the need for application specific tools and software licenses.
Restore Time: Testing shows that a restore from a known file from a disk device takes less than half the time than restoring the same file from a tape device. If the file version is known (we are looking for a file before it was impacted with a logical error), the difference is even greater. Backing up to disk also ensures a successful backup and restore, improving the success rate over typical backups to tape.
Application Availability: By moving the backup process to the network, server and storage resources are always available for production and not required to be offline for extended periods for backup purposes. SAN-based data replication can provide an instant backup copy on local disk for DR, and effectively reduce your backup window to zero. Tape backups can now be completed over the full 24-hour period instead of during a short backup window, potentially reducing the number of tape resources required and reducing costs even further.
Network-based intelligence allows you to dedicate host and storage resources to do what they do best, while improving data backup and recovery processes.
To Vitualize or Not
Some SAN-based replication solutions require that you virtualize the storage in order to replicate data. Virtualization or volume management provides value to customers requiring those features, but if you simply want to replicate your existing storage environment it may be adding additional complexity. The configuration and implementation of a virtualization-based solution is typically an intrusive and disruptive process. Generally, a virtualizing replication solution requires that all copies of the data, both source and destination, be virtualized with a proprietary header. Switching between one vendor's virtualization solution and another, or removing a product that uses virtualizing technology, is similarly disruptive. It typically involves the tedious, and error prone, manual remapping of all LUNs defined to the virtualizing platform.
However, not all SAN-based solutions use virtualization technology. In particular, the solutions that replicate to physical LUNs offer storage managers a simple installation process, plus the ability to rapidly redeploy the solution with minimal environmental impact, while providing the benefits of tiered storage architectures without mandating a specific volume management technology.
Switch Fabric Intelligence
The movement of storage intelligence, like data replication and storage virtualization off traditional server and controller-based platforms and into the fabric of the storage network, is a trend that's here to stay. And one that promises to bring enormous productivity, cost, performance, and configuration flexibility benefits to enterprise IT storage infrastructures.
As SAN-based solutions mature and this technology evolves, network intelligence will be embedded into highly distributed, next generation, intelligent director architectures that offer a variety of enterprise services, including higher scalability, more robust reliability and availability, support for multiple protocols, integrated MAN and WAN services, a variety of services for network security and quality of service, data replication and virtualization storage services, and standards-based management services, including proactive monitoring, fault isolation, and policy-based automation. This next generation of intelligent, multi-service directors will provide the fault-tolerance, dependability, performance, and flexibility required to form the foundation for storage utilities and real-time infrastructures.
SAN-based data replication offers the key to flexible, costeffective storage infrastructures and BC/DR configurations. With support for multi-vendor storage environments, SAN-based replication solutions deliver the flexibility needed to maximize the utilization of all enterprise storage assets. When combined with tiered-storage architecture to guide storage allocation and distribution, SAN-based replication demonstrably lowers storage infrastructure TCO.
WANT TO LEARN MORE ABOUT SANs?
click on: www.wwpi.com to tap into CTR's Archives
Patty Barkley is product marketing director at CNT (Minneapolis, MN)
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Storage Management; Storage area networks|
|Publication:||Computer Technology Review|
|Date:||Jun 1, 2004|
|Previous Article:||Is your company's archive really archival? What to do if you fear it isn't.|
|Next Article:||The incredible shrinking petabyte: how compression technologies are helping store more data in less space.|