Some of the more common questions around migration still relate to expected migration performance - especially for organizations migrating to Office 365.
- Will throttling impact our migration?
- What are the best methods for scaling our migration?
- What hardware is recommended to support our migration timing?
How can we tune our migration machines for optimal performance?
- What throughput can we expect?
These are critical questions for any organization planning a migration, because they affect the timing, organizational impact, and cost of the project. Understandably, these topics garner considerable attention during migration planning and are frequently discussed throughout the industry.
One migration vendor even announced this week that the latest version of their software averaged a 100% improvement in throughput when migrating from Lotus Notes to Office 365. Double the throughput should streamline the migration, right? In reality, organizations evaluating solutions and planning a migration must gather more specific information and estimates based on their needs & environment.
- What is the baseline? For example, would a 100% improvement over a 20 MB/hour/machine baseline be sufficient for your project? Percentages can be deceiving. You must confirm the total throughput numbers and architect a migration environment sufficient to meet your migration needs.
- What is the throughput estimate per migration machine? This must be derived in the environment to provide accurate data for the planning process.
- If additional throughput is required to meet the migration window, can the migration machines be tuned? Are they multi-threaded? Can they take advantage of additional CPU and RAM?
- How many machines will be needed to simultaneously migrate the anticipated number of users? How much data can be migrated during the window? What filters will be required (if any)?
Migration planning would be much easier if the answers were consistent across organizations. Unfortunately, there are many variables that impact the results including environmental factors and the migration solution & process selected.
A good starting point is asking vendors about anticipated throughput, migration architecture, and previous migration results. However, it is critical to verify the results in your environment with your data. This can be accomplished with a pilot of 20+ users and a few migration machines or virtuals. The process can be repeated with different settings to tune the environment for optimal performance. The results can then be extrapolated to establish required hardware and potential migration windows specific to your project.
This process may sound like additional work and it is. On the other hand, it is also a critical aspect of migration planning and minimizing the impact to your organization. Accurate migration estimates are prerequisites to a proper architecture and an effective migration project. The time and effort invested at the beginning will pay dividends throughput the transition.