How to plan for data migration

Computerworld - Last month, we addressed business-continuity fundamentals including data migration (see "A Business Continuity Checklist"). It's a topic that deserves its own column. In addition to business continuity, other scenarios are driving our customers to tackle data migration.
Companies are seeking well-managed storage systems to comply with regulatory requirements such as the Sarbanes-Oxley Act -- in part to combat high-cost alternatives such as distributed storage. Upgrades or migration to more commoditized platforms also present a strategic opportunity to move data.
The prospect of data migration can be overwhelming. Some of the common conditions we find among our clients are familiar to many IT managers:

• Lack of clear definition of requirements for all data. Data rules should focus on security, availability and recoverability. It's easy to imagine that documents with temporary data and permanent data could be intermingled, making it difficult to determine which data is important and which isn't.


• Distributed islands of data. Often, a business unit will implement a new application and request that the infrastructure for it remain close at hand. Unfortunately, organizational politics can exacerbate this phenomenon in the IT department, too.


• Funding constraints. Tight budgets may limit technology decisions and options, or a company may invest in technology for projected bottom-line benefits -- only for other factors to interfere with hoped-for business impact.


• Lack of expertise in heterogeneous storage environments. With each vendor's support limited to its own products, incompatibility between storage technologies becomes the problem of the IT manager.

1. Detail security and availability requirements. Sometimes called data classification, this requires the security and infrastructure teams to jointly identify the needs of the IT environment and ways in which data may be segregated and protected. Data classification describes conditions for data access, retention requirements and security measures such as encryption. Even a limited set of classifications will have tremendous benefit.


2. Define migration requirements -- which determine success criteria. These may include new service-level agreements, expectations for the new storage infrastructure, and objectives such as reduced management costs, reduced storage expenditures, greater insight into expenditure, a simplified vendor model or greater technical flexibility or stability.


3. Survey the IT environment. IT departments often use tools and scripts to do this. But migration requires a complete understanding of infrastructure technology involved, including the networks and file servers. The location of data is just as important: Without providing sufficient bandwidth capacity to support heavy network access, relocating data to a centralized repository could have adverse effects. A survey should detail the following:
   
   • The location of the data, its capacity and growth requirements.
   
   
   • Data usage as a measurement of the I/O load, sometimes called an I/O characterization.
   
   
   • Data importance, which can reveal potential effect on network load and influence retention and availability requirements.
   
   
   • Data classification. IT managers can decide what data requires the most expensive storage and stringent protection and what could be re-created or restored easily from an archive. They can then make well-informed, strategic decisions about future systems.
   
   
   • Management costs for the current environment. This often-overlooked step offers the best opportunity to define benefits of migration.
   
   
4. Design the appropriate consolidation or replacement platform, including technology, management and backup tools and procedures. Data classification and a good survey reduce the chance of an overengineered system and contribute to a platform design that will accommodate growth, availability and performance. Plus, a consistent management environment can trim the most expensive aspect of the storage environment, administration.