Data Migration: Old Headache, New Aspirin

Send to friend

In a recent data migration study I was mildly surprised to read that 85 percent of migration projects fall short of their goals because of problems with technical compatibility, unplanned downtime, and data loss or corruption. This statistic is somewhat depressing because migration has been around as long as the IT industry has been with us. One would think by now customers would demand, or vendors would have delivered solutions alleviating the pain. I say mildly surprised though. In the open systems quest for price performance, distributed complexity in IT has grown substantially; now IT departments are looking to consolidate back to a reasonable balance.

Storage vendors, looking to protect their installed base, have by-and-large offered proprietary migration solutions. This has left a void in the market that is being filled by ISVs with more flexible server-based offerings. These software-only approaches can reduce project times and compatibility hassles of migration across heterogeneous environments. When considering these alternatives however, three criteria should be kept in mind--- platform coverage, network utilisation, and recovery performance.

In general, the broader the platform coverage the better, but make sure you inspect vendors prerequisites. Some require specific application software, file systems or volume managers; make sure the solution supports different storage architectures as well, define storage source and target mappings based on all required data for application recovery. For example Exchange application data can be SAN resident, but system disk data may be on the server internal drive--- both are required for quick recovery.

Time to synchronise source with target data over the network can be prohibitive; so plan to use data compression/reduction. These features should be a standard solution component and can give 3-10x improvement in network load. But beware of vendor claims; performance varies widely depending on disk write activity. Also, consider offline initial synchronisation (OLS). This nifty feature is supported by a few enterprise-class offerings. With OLS you create a baseline point-in-time copy and ship it to the target site, where it is merged with subsequently replicated I/Os, creating a current, consistent image. This bypasses network synchronisation saving time and expense.

No matter what your strategy for coverage and network utilisation, if you cant quickly recover data and seamlessly cut-over, interest in your project will increase exponentially from all quarters. First, to help avoid this problem, make sure your migration solution can recover from outages during replication, still maintaining target data consistency. You dont want a surprise at the finish line, and you dont want to start over every time theres a transient system or network outage during migration. Secondly, test before cut-over. Some solutions support testing in parallel with migration without production downtime. This step can save countless headaches at cut-over. Lastly, make sure you can recover applications with dependent data on multiple servers. For example, many companies run Exchange on one server and the related Active Directory on another. These and other distributed applications such as SAP demand recovery of multiple environments to the same point-in-time. For these requirements, solutions should be evaluated with care as to their capabilities.

Data migration can be complex, time-consuming, and costly. However, alternative technologies are emerging to challenge the incumbents. With careful upfront evaluation and planning these solutions can shave substantial cost, time and risk off data migration projects.


Chris Hyrne, VP marketing, is responsible for product management, product marketing and communications worldwide. He brings over 20 years experience to Topio with a successful track record in senior management positions for marketing, business development and sales in the high technology industry.

Topio provides software for data replication and recovery across the spectrum of locations, platforms and storage that support your enterprise. Worldwide, companies of all sizes depend on Topio for: disaster recovery, backup consolidation, and data migration.

Comments (0)

Add a Comment

This thread has been closed from taking new comments.