Jan 12, 2023
If you’ve worked in and around enterprise IT for any length of time, you’ve been part of a project to move data center resources from one location to another. Whether it’s a corporate relocation, an acquisition, merger, or divestiture, very few network engineers avoid the dreaded data center migration.
Your approach to a data center move will vary depending on several factors, including your familiarity with the environment, your organization’s security posture, the current technology in use, and the desired end state for the migration.
Let’s step through a few factors you need to consider as you begin planning your data center migration.
If you’re migrating an acquired data center, you’ll almost certainly encounter overlapping RFC 1918 IP address space. There are only so many ways to deal with this problem. From a networking perspective, the best solution is to re-IP resources that are using duplicate IP addresses. If you’re able to re-IP resources, you can develop a IP scheme that works with your current infrastructure. As an added benefit, a re-IP effort helps your organization understand application dependency mappings and may reveal issues that need to be addressed before a migration.
Unfortunately, few organizations understand their security and application dependencies well enough to embark on a large-scale re-IP effort. Combine this reality with a tight timeline and you may not be able to convince leadership to re-IP. In this case, it’s helpful to think of the data center you’re migrating as a separate tenant. Work closely with your virtualization team to examine the requirements for multi-tenancy. Smart investments in the right technology in this arena can save hundreds or thousands of engineering hours.
Regardless of the technology you choose, some combination of tunneling (VXLAN, GRE, IPSEC) and NAT will be required to move resources.
When migrating resources from one data center to another, you need all the connectivity you can get. Research your options and be sure to consider redundancy. Be clear with your leadership about the costs and potential impacts of varying connectivity options. Understand your data migration needs, including the data change rate. Use a bandwidth calculator that takes into account latency to determine how long data transfer and continuous synchronization will take.
Your use case may justify SD-WAN technology to make the most of the connectivity you have. Communicate clearly with other teams about their expectations related to data transfer rates. Many server admins have wildly miscalculated data transfer rates because they assumed data transfer rates were in bytes, not bits, or they didn’t account for the impacts of latency and TCP handshake on data transmission rates.
As you develop a timeline for moving resources from one data center to another, there are several factors to take into account. Do you need the ability to move resources a VM at a time, while maintaining the IP address? Do you want to move resources VLAN by VLAN? Do you need to keep all the tiers of an application together because of latency concerns? Are you going to attempt a Big Bang approach, where you move everything at once?
As you think through each of these methodologies, be sure to consider Internet access requirements, security dependencies, and user access needs. Remember that as resources move from one data center to another, the source public IP address of that resource will likely change as traffic egresses a new environment. Be sure to notify third parties that may restrict access to their services by source IP.
Depending on your use case, IaaS services in the cloud may become a helpful tool in your tool bag. If you need to migrate resources out of one physical location on a tight timeline, but have work to do before integrating those resources into your on-premises environment, you may want to consider migrating resources to the cloud. Once your VMs are in the cloud, you can re-IP, scan, patch, secure, and update them. You can migrate applications or services to your on-premises infrastructure and then spin down the cloud resource when you’re done with it.
Regardless of the choices you make, successful data center migration requires an integrated IT team that considers requirements carefully, documents diligently, plan well, and executes with precision. As a network engineer, your secret weapons will be research, attention to detail, the right tools, and strong communication skills. Combine these efforts with a little luck for a successful data center migration weekend.