Over the past decade, cloud services have rapidly become one of the most defining technologies in IT. The hype surrounding cloud services may make it seem like all of an organization's resources should be migrated to the cloud immediately. There is no denying that, in certain cases, cloud services can be tremendously beneficial. In others, however, a cloud migration probably doesn't make sense.
Organizations need to take a hard look at their existing investments in infrastructure -- from hardware to application portfolios to network architecture and beyond -- to determine if a move will be beneficial. Some of the migration questions are technical, such as whether a given application can perform adequately in the cloud; some questions will involve nontechnical, budgetary issues, such as whether a cloud migration is cost-effective given current investments in infrastructure.
Here we explore factors that should guide a cloud migration strategy and help determine whether to move on-premises workloads to the cloud.
One of the first considerations is an organization's existing data center investment. Despite technologies such as server virtualization, there are real costs associated with deploying on-premises servers. There are not only licensing costs involved, but also costs associated with hardware resource consumption and support infrastructure. As such, there is almost always a significant investment associated with an on-premises server. Outsourcing a server's data and/or functionality to the cloud may mean abandoning your on-premises investment unless an on-premises server can be repurposed.
Although this rip-and-replace approach to cloud migrations may not make financial sense for organizations that have a large investment in an on-premises data center, an organization can still benefit from migrating certain on-premises resources to the cloud.
No matter how good it is, any server hardware eventually becomes obsolete. Enterprise-class organizations have traditionally coped with this expected obsolescence by adopting a hardware lifecycle policy. An organization, for example, might choose to retire servers after five years. That being said, an organization could integrate a cloud services roadmap into its hardware lifecycle policy. Doing so allows IT teams to migrate on-premises resources to the cloud instead of moving them to newer hardware.
The prospect of using cloud services is often particularly attractive for smaller organizations and startups. In the case of a smaller organization, the use of cloud services provides access to enterprise-class hardware and fault-tolerant features that would otherwise be unaffordable. Similarly, startups can benefit from cloud services because they can get their operations running quickly without having to invest in on-premises data center resources.
Application requirements for a cloud migration
In the case of application servers, administrators must consider whether the application can function in the cloud. Likewise, the application's performance must be considered.
Compatibility usually isn't a big problem for newer applications that run on top of modern operating systems. It is also easy to assume that performance won't be an issue for such applications because most cloud providers will allow hardware resources to be allocated to hosted servers on an as-needed basis. However, two major considerations must be taken into account for such applications.
The first is performance. Even though you can provision the hosted application server with nearly unlimited compute and memory resources, Internet bandwidth may impede application performance. It does little good to have a high-performance hosted application server if Internet bandwidth limitations stand in the way of a good user experience.
The second consideration is application portability. Although it is often easy to migrate a virtualized application server to the cloud, the application might have external dependencies that rule out (or greatly complicate) a cloud migration. For example, the application might have an Active Directory dependency or require access to an on-premises SQL server database.
For older applications that run on legacy operating systems, a move to the cloud may not be an option. Lab testing is the only way to know how an application will behave in a cloud environment. Testing helps determine the steps that are involved in moving the app there.
Another consideration for moving application servers to the cloud is hardware scalability. Some IT analysts have suggested that cloud services are ideal for hosting hardware-intensive workloads because cloud services generally offer nearly unlimited scalability. While a cloud service provider can usually scale its offerings to meet even the most demanding workloads, this scalability comes at a price.
Infrastructure as a Service providers such as Microsoft and Amazon Web Services charge customers a resource consumption-based monthly fee. As such, a cloud-based high-performance computing environment can become cost-prohibitive. Recently a client told me, for example, that it costs more than $10,000 per month to operate a single high-performance application in the cloud. The bulk of the cost is due to CPU and disk I/O consumption.
Virtualization will ease a cloud migration
Regardless of organizational size, one of the considerations is whether the workloads targeted for cloud migration have been virtualized. In some cases, it's much easier to move workloads to the cloud if on-premises servers have already been virtualized. In fact, some providers will allow an organization to port virtual machines directly to the cloud. If on-premises servers have not been virtualized, a migration to the cloud is likely still possible, but the process may involve more work.
Cloud infrastructure considerations
Another factor to consider is the on-premises network. If an organization plans to keep resources on-premises (even temporarily), the cloud network must function as an extension of the on-premises Active Directory forest. This means that the organization will typically have to deploy cloud-based domain controllers, DNS servers and possibly DHCP servers. More importantly, the organization will have to figure out how to establish a secure communications path between the cloud-based virtual network and the on-premises network.
This requirement usually isn't a deal-breaker for organizations with an existing on-premises network, but it does mean that a significant amount of planning may be required before beginning the migration process.
As an organization contemplates the risks and benefits of cloud migration, it is important to keep in mind that cloud migrations are not an all-or-nothing proposition. Organizations do not have to go "all in" with cloud migrations. In most cases, it will make sense to move certain services to the cloud while continuing to operate others on-premises.
About the author:
Brien Posey is a 10-time Microsoft MVP with two decades of IT experience. Before becoming a freelance technical writer, Brien worked as a CIO for a national chain of hospitals and healthcare facilities. He has also served as a network administrator for some of the nation's largest insurance companies and for the Department of Defense at Fort Knox.