Data protection and quick recovery are critical requirements for enterprises that have important business data growing at a fast pace within their data centers or remote locations. Here are some options for data protection and recovery architectures in distributed and centralized storage configurations. by Anal Jain

Storage growth and investment on storage solutions within enterprises is increasing at a fast pace. As businesses depend on IT infrastructure for higher productivity and competitive advantage, data protection and recovery become critical for everyday business operations, especially during major failures or natural disasters. This highlights the need for data replication, backup, and restoration of operations, based on optimal utilization of IT resources.

Today, customers are considering various approaches for quick recovery to protect their valuable data and mechanisms. Some of the factors which affect these infrastructure and policy decisions are cost of downtime, backup windows, time-to-recover, frequency of backups, and maturity of technological options. Such decisions typically result in a combination of online data replication (disk-to-disk) or offline (disk-to-tape) backups and the implementation of policies required with these operations.

Let's focus on data protection strategies and for networked storage in both distributed and centralized topologies based on Ethernet/IP network infrastructure.

Variety of features
Networked storage solutions should offer a variety of features that address high availability and resilience for data storage like clustered configurations, RAID protection, and robust product architecture. The network infrastructure supporting these storage topologies also enhances the overall resilience of the solutions in terms of high availability and security.

The key factor that drives the stringent requirements for data availability is the cost associated with downtime. This includes costs associated with lost productivity, reduced customer satisfaction, and the opportunity cost from lost revenues due to inability to access business-critical data.

Accidental data deletions and data corruption by an application may lead to irrecoverable losses resulting in significant effort and time expended to reconstruct the data. To ensure data recovery from such catastrophes, most IT organizations archive or backup to a relatively inexpensive offline media, like tape. In most cases, the periodic backup process is performed directly off the production data, rendering it inaccessible to users or applications during the backup cycle. This is due to inherent limitations in the data transfer speed associated with such media. Similarly, recovery of data from such archives also leads to significant downtime.

Natural disasters like earthquakes, floods, and fire can take the entire IT infrastructure down. In such cases, reconstructing all the business-critical data and rapidly recovering to full business operation can be extremely difficult. Such data recovery could take days, which could severely affect the functioning and viability of an enterprise. Many far-sighted enterprises implement disaster recovery plans to guard against this potential calamity.

A disaster recovery implementation typically involves copying data from the production site to an online 'disaster recovery' site. Such an implementation involves not only infrastructure, but also people and process-related design considerations.

Disk-to-Disk-to-Tape Deployment
A new breed of networked storage technologies called Nearline, can provide a unique approach for LAN-based backup. Nearline supports cost-effective disk shelves with the purpose of providing faster disk-to-disk backup mechanisms. Mirroring software can be used for asynchronous data replication over an IP/Ethernet connection from networked storage at remote sites to a Nearline appliance located at the central site.

Online Data Protection Solutions
As the nature of data becomes more critical and the cost of downtime increases, customers need to consider solutions that allow data protection and recovery from an online repository in addition to archiving to tape. The cost of downtime in terms of its business impact and customer satisfaction issues justifies these solutions and topologies. The following section describes how networked storage can be deployed to provide online data protection, both within a data center as well as in distributed environments.

In terms of online data protection, let's review the relevant features that should be part of a networked storage solution. The 'snapshot' feature enables users to create instant online copies of data. Snapshots use a copy-on-write technique to avoid duplicating disk blocks that are the same in a snapshot as in the active file system. Only when blocks in the active file system are modified or removed do snapshots containing those blocks begin to consume disk space. There is no performance impact due to snapshots taken at any frequency of intervals. And the networked storage solution has a built-in software feature in order to recover from data loss.

Remote Site Disaster Recovery
The solutions described so far can only address data protection and recovery within a data center or campus environment. However, to protect against a data center or site going down, as a result of events like earthquakes and sabotage, customers should consider a more cost-effective remote disaster recovery implementation.

The configuration options for remote site disaster recovery vary widely depending on the distances between the sites, level of redundancy required, and other policies for data recovery.

Active/passive configuration
In this type of configuration, storage at the primary site uses mirroring technology to provide frequent incremental updates of the data to the storage at secondary site, which acts as the online backup repository for the primary site.

Active/active configuration
The active/active disaster recovery configuration is similar to the active/passive configuration except that the secondary site is also a production site. Just as the primary site data is replicated to the secondary site, the secondary data is also mirrored back into the primary site with the help of mirroring software for dual protection. This allows either site to recover in case of a disaster at the other site. Each site also continues to serve data to local clients.

WAN configurations
In case of sites that are distributed over distances larger than a campus or metro area, the backup configuration requires the use of WAN links connecting the sites with each other. These topologies highlight parameters like QoS, security, cost of the WAN link, and whether the link is dedicated for storage or is shared with other types of voice or data traffic.

Security is always considered an important design parameter. The topology might include IP firewalls, VPNs, or IPSec, which act at the network layer, protecting and authenticating IP packets between participating IPSec devices. These design components may also require more consideration in terms of application latency and response times.

Mirroring technology is highly efficient in its bandwidth usage over the WAN links. The design involves transferring only the incremental data at user-defined intervals from the target storage to a Nearline storage device. The incremental data is now archived to shared tape-libraries.

If the WAN link is shared with other types of real-time or latency-sensitive traffic, it is imperative to consider network QoS at the network layer and look out for limitations like throttle in the mirroring software.

Anal Jain, MD, India and SAARC, Network Appliance can be reached at anal.jain@netapp.com