Many factors drive the need to migrate data. Storage systems may be upgraded for better performance and capacity or replaced when they come off lease. Applications may need to move from test or development to production systems. Disk distribution, redundant array of independent disks (RAID) groups and other factors that affect where data is placed are reorganized periodically as performance or protection needs change over time.
In addition, with data stores growing exponentially and server and storage resources undergoing virtualization and consolidation in data centers throughout all industries, the management of data mobility to optimize operational efficiency is an ongoing challenge that has the potential to consume IT time, resources and budgets. Which is why IT managers need data migration tools that enable data mobility by providing low-impact, automated, host-free movement of data across storage networks and -- with the advent of cloud storage -- across the wide area network (WAN) into the cloud.
Despite the pervasive need for data migration in IT operations, migrating storage area networks (SAN)-based data to different storage systems is an ongoing challenge. Disk array tools are normally incompatible across different vendor systems. Host-based tools are intrusive and disruptive and impact application processing. Moving data using backup and restore from tape is slow and cumbersome and does not scale well. The result is that data migration projects tend to be time-consuming and costly, requiring extensive application downtime, affecting business performance and productivity.
In the manufacturing industry, the need for a non-disruptive data migration process is critical, as manufacturing processes cannot be suspended for very long. There are many tools out there today that facilitate data migration in 24/7 environments and across multiple sites, so here are a few of the critical attributes to look for when choosing the right migration tool for your operations.
Challenges in performing data migrations
In performing any data migration, you will be up against a host of obstacles and challenges. First of all, there is never enough time for a clear window during which you can suspend applications for IT maintenance. Most manufacturing operations need to run at peak performance with little or no downtime, as any downtime costs money. In addition, most environments are an amalgam of products from different vendors, so both the existing storage resources and the new storage to which you are migrating may be a mixture of different products from different vendors. Finally, on a more granular level, there are different logical unit number (LUN) sizes and types to accommodate, perhaps silos of old direct attached storage to encompass, and often little or no centralized management.
Elements of a useful data migration tool
There are a few key elements that a decent data migration tool must have to help you migrate data when and where you need it.
- Look for a system that is fast, reliable and non-disruptive to operations.
- Make sure it is entirely storage based, so that it can work independently of host applications and operating systems and have no impact on daily operations.
- It must support all disk systems, across all vendor boundaries.
- And, lastly, it should not alter or reformat your data in any way, so that what you get is a precise, block-for-block mirror of the original data -- ensuring that data integrity is maintained.
How data migration works
The best data migration tools will tap into your network and have the ability to recognize all available data stores, regardless of the heterogeneity of your storage systems. This requires a level of sophistication that is able to virtualize different logical devices and recognize them all in one consolidated view. You need a data migration tool that will see whatever storage is present and be able to work with it.
The tool has to connect to the existing storage LUNs without the need to modify the native data format. Once this type of system is in place, the new target storage is connected to the SAN and is also zoned to the data migration device, so that it can act as an intermediary and traffic controller for the data transfer. With the target storage in place, the existing data LUNs are fully migrated to the new array. The abstraction layer will allow you to move data to the new storage resource even if it is from a different vendor, uses different drive sizes, or uses different RAID configurations.
The best tools will provide block-level, synchronous mirroring between the two storage frames. There is no need for application downtime during this process. To ensure that primary application processing is not adversely impacted during mirroring, you will want your data migration tool to offer extremely granular controls over quality of service (QoS), input/output (I/O) throughput, and so on.
To complete the process, as soon as the mirror is fully synchronized, you should be able to reverse the mirror priority through a simple push of a button. In other words, the original storage is re-defined as the mirror, while the new storage is defined as the primary. At this point, the mirror can be broken, leaving the application running entirely on the new storage. Data processing continues uninterrupted throughout this process and the old storage systems are seamlessly removed.
Clay Burns is vice president of professional services of FalconStor Software, a provider of data protection solutions, where he is responsible for assisting North American companies in designing, developing, implementing and testing storage solutions. For more information, visit http://www.falconstor.com/