(Un)Expected Journeys on a Cloud Migration
By Scott Chate, VP Partner & Market Development, Corent Technology
Cloud migration is a journey, and like the migrations of animals, it’s usually done more than once in a lifetime. Many of the organizations who migrate their applications to the cloud may have previously done migrations from one data center to another or from one technology to another. Each of those types of migrations has unique challenges. So, the migration to the cloud pose unique challenges.
Some of those challenges will be expected, and some will likely not. Migration to the cloud can involve deeply technical considerations ranging from hypervisor availability to SAN and security of networking, to ordinary considerations of cost and ability to transfer volumes of data to the cloud.
There are two primary ways to approach a cloud migration, wholesale shift of a data center, and application specific migration. Depending on the situation and goals of the migration either one might be the appropriate choice. Often it can be a hybrid, with some servers being migrated and in other cases some specific applications being migrated.
"For Application oriented migrations, the focus is on getting a particular application to function on the cloud"
A data center migration is often concerned with deadlines related to a contract for a data center or for the hardware that is ending. The goal is typically to get everything moved as quickly, cheaply and problem free as possible. This type of migration is most akin to the traditional hardware upgrade or data center change cycle, although doing every application in the portfolio at once can add complications.
In this type of migration, stability is a primary goal and so a technique that uses the exact image of the source server is preferable. A migration capability that allows for an automated scan and analysis of the data center (or specified sub-set of it) is essential to collect the necessary information.
What may not be expected is the work to adjust the hypervisor to one compatible with the cloud, the need to adjust VM sizing on the cloud to reflect the performance of cloud virtual machines in comparison with existing hardware. Simply having the same number of virtual CPU’s or memory allocation in the cloud may not be enough power to equal the hardware.
Once the source servers are identified, custom empty VM’s from the cloud, ones that don’t have any OS on them, can be used to provide the compute platform that replicates the environment. You must have the ability for the migration tool to make adjustments as it provisions the cloud IaaS resources.
And with the new cloud, VM’s their may be networking and storage configurations that have to be done in a cloud compatible manner and replicate the functional characteristics that were imbedded in the original data center. The ability for the migration tool to run the appropriate configuration scripts, and at the appropriate times, is essential to a smooth transition.
For Application oriented migrations, the focus is on getting a particular application to function on the cloud. Often this is combined with a desire to improve performance and scalability and lower the costs of operating the application. This requires a much smarter approach to shifting the application.
To utilize the highly touted advantages of the cloud such as elasticity and scalability, the fact that it’s essential to perform a migration that allows for a change in the topology of an application may be unexpected.
The clouds has scalability and elasticity, but only if the application is capable of taking advantage of it, and unfortunately many are not deployed in an architecture that can support those cloud capabilities. Needing involvement of an architect capable of analyzing the topology and components of an application, and designing new topologies that provide the opportunity to utilize the cloud effectively, is often an unexpected burden. Having a migration tool capability that provides both analysis of an application topology and components, and that can generate options based on goals such as higher performance, availability or scalability, and cost control, can provide an unexpected bonus to the migration.
Cost is a major factor to consider when migrating to the cloud. A migration tool capability that allows you to automatically compare projected cloud costs, either across different clouds, or using different deployment topology models, provides a way to avoid unexpected surprises. Being able to compare costs for different cloud deployment plans can often reveal unexpected good news. Sometimes implementing a scalable, elastic topology model with right-sized VM’s will allow a better performing application environment with overall lower costs by allowing dynamic changes or substitution of PaaS services for some application functions. Normally this requires extensive work in application development, but if the cloud migration capability can implement these features automatically, then there is an opportunity to get a higher ROI for the migration.
One thing that is constant in IT is change. You are likely to do another migration in the future. It could be to migrate more applications, to move to a better cloud where price, performance or reliability is better, or simply because the regulatory environment mandates that some customers must be managed on clouds that meet certain criteria. The ability to migrate again, either to another cloud, or another data center of that cloud, or to a new topology that will enhance the application further is essential in a migration tool. Having the ability to migrate to a more optimal application topology, on the same or a different cloud provides flexibility to adjust with demands as they evolve.
This ability to continuously evolve the application deployment topology and environment is essential to achieving the benefits we’ve all been told to expect from the cloud.
Established in the year 2000, Corent Technology, a California based company offers services specializing in Software-as-a-Service (SaaS), Cloud computing, Cloud, Platform-as-a-Service (PaaS), Enterprise software, SaaS appliction development and more.