Your primary concern when migrating data centers is making sure services remain available. Learn how to approach this issue, as well as how to migrate hardware and data safely.
Data center migrations are complex operations that can be difficult to explain to executives who write the check for the migration activities and need to understand and manage the associated business operations risks. We'll take a look at some of the complexities and risks associated with migrating a data center.
The primary purpose of the data center is to host applications that service the business. Whenever you consider migration from one data center to another, you must first consider the availability of the underlying services. These services include infrastructure application such as Active Directory and client-facing applications such as SAP.
As services shift from one data center to another, you must form a strategy that factors in when specific services get moved and the interdependency of applications on one another. A common approach to addressing service availability is to develop move groups and then place interdependent applications in common groupings.
For services that support most enterprise applications such as Active Directory and DNS, a common approach is to extend these core services across the data center. The services remain in both data centers until the migration is complete.
There are two strategies to migrating physical servers: one is known as "lift and shift" and the other is data replication. In a lift and shift strategy, the hardware is put on a moving truck and installed in the new data center. The system is backed up prior to relocation, but there are risks associated with this strategy.
One of the largest risks is damage to the physical hardware during shipment; damage during shipment can render backups useless. Another challenge is the physical distance between data centers may not permit this option and have services available within an acceptable period.
The second strategy is to perform data migration over a leased circuit. With a leased circuit comes two sub-swing hardware options. One option is to perform a physical to physical (P2P) migration. A P2P migration involves acquiring like hardware that both the application and hardware can be migrated to while keeping downtime to a minimum.
The other hardware migration option is a physical to virtual (P2V) conversion. A P2V involves converting a physical machine to a virtual machine over the leased line. P2Vs serve two purposes: the first is to migrate a workload from one data center to another while keeping hardware costs to a minimum; the second is to undertake a data center transformation by moving to a virtual platform. P2V migrations are popular options, as many engineers are already accustomed to performing these conversions as part of previous data center projects.
Getting application data from one site to another may be one of the most complex parts of a data center migration. A simple option would be to perform a tape- or hard drive-based backup and perform a restore; however, similar to a lift and shift migration, backup and restore provides limited capability for restoring services in a timely fashion. Also, backup and restore isn't an optimal method for data migration -- it's better suited for disaster recovery where data recovery options are limited.
The primary method chosen for most data migrations is the provisioning of a leased line. With a dedicated connection between data centers, a migration team can leverage hardware- or software-based synchronization to perform the data migration. Along with the ability to migrate data, this method can be leveraged to perform P2P, P2V, and virtual to virtual (V2V) migrations.
Many organizations choose to have multiple connections between data centers. Connectivity involves a minimum of two circuits: One connection supports regular end user and data center to data center traffic to support applications such as Active Directory and application to application traffic, and a second and normally faster connection is used to perform data synchronization. The dual connections keep the two very different traffic types from interfering with one another.
A successful data center migration strategy begins with identifying the business around availability prior to any migration activity. Technical solutions fall into simple categories after requirements are gathered.
What data center migration risks would you add to this list? How do you recommend handling them? Share your experiences in the discussion.