Keep data migration costs low with swing or rental hardware strategies
Image: Tomasz Wyszołmirski
One of the major challenges in migrating to a new data center is having a landing spot for your workloads and data during the transition. In the best case scenario, you are moving to a green-field where all new hardware will host your existing workloads, and the only concern is the actual migration work. However, green-field opportunities are rare; in most cases, data center managers have to develop swing hardware strategies, which minimize wasted hardware and keep costs in control. Here's a look at common hardware procurement options and considerations.
SEE: 3 risk factors and strategies when managing data center migrations
Swing hardware
A popular option is to buy a small footprint to use as swing hardware. The concept is to purchase the bare minimum to move less critical workloads at the beginning of the migration.
For example, your largest workload may be an Exchange server with 32 GB of RAM and 1 TB of RAID 5 storage. An example workflow follows.
Migrate small workloads to the large server in the new data center.
Relocate the original hardware.
Migrate the workload a second time to the original hardware.
The challenge with this approach is having enough of the right kinds of hardware during each migration wave. Also, the amount of work and risk is increased as the workloads migrate more than once. If you are not in a heterogeneous hardware environment, this may not be a very good option; there may not be enough of a single type of server to make this approach cost effective.
This strategy also only takes server equipment into consideration. If you have a larger environment that includes complex networking as well as an enterprise SAN array, the swing approach costs increase dramatically.
Rental hardware
Another option is to rent a subset or the entire footprint of your destination environment. To limit the impact of multiple migrations, many enterprise vendors support a rent and replace strategy. In a rent and replace strategy, the vendor will rent your existing hardware profile and accept your original equipment in return. From a cost perspective, this may be close to the cost of purchasing a subset of your environment and using a swing approach, though the migration work is reduced and with fewer service outages and reduced risk. An additional advantage to this hardware procurement strategy is that network and storage become an option as well.
If you are going into a hosted data center, there may be an option to rent the hardware directly from the hosted data center provider. Be prepared to have limited options, and don't expect to be able to rent exact models. This option will vary widely between data center providers.
This option isn't without challenges or risks. One challenge is software licensing. In many environments, some software licenses such as OS and virtualization platforms are tied to the hardware; these licenses don't automatically transfer to rental hardware post migration. Support is another challenge. More than likely the refurbished rental equipment is from refurbished stock; you would have to negotiate support as part of the rental terms of the hardware. One risk is costs associated with migration delays. These rental agreements are normally month-to-month. If the migration is delayed, the hardware procurement costs increase due to an extended rental period.
Conclusion
Data center migrations are risky undertakings with many unknowns. The hardware procurement strategy you select can either ease or compound the challenges. You might be surprised by the options hardware and data center providers have for you to consider.
