by Thomas Crowe (Red Hat)
As an experienced IT professional, chances are you’ve been involved with a migration of some sort. Whether it’s a simple migration, such as moving static data to another node or a highly complex migration across datacenters, all successful migrations have one thing in common – rock solid planning. Migrations that are attempted without the requisite planning can be fraught with peril, and end up with disastrous consequences
Ultimately, users, our customers, do not really care if a given server is up or down. They care whether they can access a specific application, such as email, a web site, or data. It is the service that users care about, and it is the service in which migration planning needs to be focused.
Historically, migration planning has been very tactical, revolving around the hardware, the operating system, or the application. IT targeted whatever needed to be upgraded and simply replaced it with a newer or better one. When applications were monolithic and IT environments were homogenous, this approach worked. Today, services have evolved and are generally comprised of numerous applications hosted on various platforms across multiple locations. This increasing complexity unravels the old approach and forces a move away from hardware-centric, OS-centric, or application-centric, to a holistic, services-centric focus. And, it is this move that shifts us from tactical to strategic migration planning.
Strategic Migration Planning focuses on workloads and interfaces as the driving considerations. The key thing to keep in mind about a workload is that users see it as a single service despite the number of applications that comprise it. With an understanding of the workload’s applications and their interfaces, monitoring can be put in place to ensure that every system accessing every interface is documented and accounted for in the migration plan and the corresponding migration test plan.
Strategic Migration Planning exposes several key pieces of information that can be more easily determined than with the traditional tactical approach. For example, by understanding the underlying business function(s) or workloads that an application supports, the required SLA can be more easily determined. Also by examining workloads, application interactions are more readily apparent and testing scenarios can be developed to evaluate each of those interactions.
Strategic migrations are more complex and require significantly more information than ad-hoc, tactical migrations. Fortunately, there are tools available to greatly lessen the burden of collecting that information. The most significant could easily be a robust enterprise Configuration Management Database (CMDB). By having a reliable, up to date, CMDB, onerous data collection turns into simple queries. Additionally, all of the data, or a significant portion of it, become readily available thereby reducing the planning and documentation effort required for a successful migration plan. Those organizations that have a strong CMDB practice generally are more successful in managing migrations. Those that don’t, should.
Planning migrations based on services rather than servers is somewhat of a paradigm shift for many engineers and administrators; however it is one that that is necessary and ultimately more efficient given the increasing complexity of todays IT environment. Ultimately the dividends received are substantial. Smoother, more reliable, and effective migrations; with fewer back-outs and unexpected problems are just a few of the benefits of a strategic migration plan.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.