6 Tips To Mapping Out Cloud Migrations12:00 PM EST Wed. Jun. 05, 2013
It seems like you can't swing your arms nowadays without hitting some company in the process of moving at least part of their IT resources to the cloud. But aside from satisfying customers' high-tech fashion needs (OK, there's that cost thing, too!), such moves also require a great deal of planning, as well as technical expertise. Rackspace Hosting has come up with a list of things that definitely belong on your to-do list when planning such a move. CRN has taken a few liberties with Rackspace's advice, based on the ongoing feedback we hear from the channel.
Much like a workday commute, the preparatory phases leading to the big arrival can often run into traffic jams, and even the occasional logjam -- either of which tend to throw a wrench in even the most carefully crafted timetable. Rackspace recommends maximizing the amount of heavy lifting before the cut-over. And don't forget to change the DNS for your domain, and, in most cases, to set the hop limit as low as possible. The company also recommends allowing extra time for the DNS to propagate, given that the process often takes longer than expected.
Going into the experience thinking that an app is an app is begging for trouble. Some apps are honest-to-goodness cloud apps that were designed for the cloud from the ground up. Others are based on traditional software packages that were later converted for cloud use, in some cases more successfully than in others. Rackspace recommends ascertaining how the application is architected and what types of bottlenecks or other issues are being experienced in the wild. It also advises focusing on which portions of the app carry the heaviest loads because that information will help to ascertain the level of scalability. It's also important to ensure the app "can assume multiple end points on each different layer (e.g. web, application and database)," according to Rackspace.
Testing the performance of the application after the migration is a well-known part of the process, but Rackspace points out that similar performance testing can be done in advance of the cut-over. The company explains that most applications being deployed to the public cloud will have multiple endpoints, so they recommend repeatedly monitoring the performance of the hosts. Rackspace points to its own cloud monitoring service as a means of achieving this objective, but its blog entry also says that alternative open source monitoring solutions, such as Nagios, can also be effective.
Various industries and political boundaries can carry a variety of compliance restrictions that can sometimes be difficult to track in the midst of seemingly more pressing concerns, such as closing the deal. The good news is that most channel partners are not strangers to compliance requirements, such as PCI and HIPAA. But Rackspace reminds us that some of the restrictions might only apply to certain layers of the configuration. The blog cites an example in which an e-commerce business hosts the PCI compliant layer on dedicated servers, which then connect to a web layer hosted in the open cloud.
Nobody likes to consider the possibility that they might need to undo what has been done because something went seriously wrong. But when dealing with complex functions such as cloud migrations, it's important to identify in advance the processes for reinstating the previous configurations. Hopefully, such contingency plans will never need to be activated, but Rackspace warns that failure to be ready can have worse implications than a temporary retreat.
The cloud is all about efficiency, so Rackspace advises that customers use APIs -- especially theirs -- as a means of maximizing that benefit. The company says that integrating applications with its API can support more extensive automation for functions such as horizontal scale-out of the configuration as loads increase.