Mathias Rosenthal - Fotolia
With companies such as Amazon Web Services, Google, Microsoft and Rackspace offering the ability to create virtual machines in the cloud to support and replace physical servers, cloud virtualization services are being integrated into data center infrastructures. But knowing which features to consider and which vendors to compare can be a daunting task.
This article is the first in a series that walks through the buying process for public cloud virtual server services. Here we cover the major benefits of using public cloud services as an extension of your data center infrastructure.
The second article discusses the risks and costs associated with moving virtual machines (VMs) to the public cloud. The third article focuses on purchasing criteria and preparing a vendor request for proposal (RFP). Finally, the series will compare market-leading services against established criteria and against each other to help you select the best public cloud service for your environment.
In most cases, established organizations with IT resources on-premises should not dispose of existing servers and move everything to the cloud. It would be a waste of money, unless the local resources were scheduled to be retired. Even then, there may be some workloads, such as latency-sensitive ones, that should run locally. Similarly, you might not want to put all of your domain controllers on a public cloud.
Of course, this does not mean companies with established IT infrastructures cannot benefit from the cloud. The best approach is often to treat the cloud as an extension to the organization's existing IT footprint. In this feature, we look at the benefits of using public cloud versus an on-premises data center. In the second part of this feature, we'll cover the disadvantages of public cloud.
There are multiple ways a data center benefits from being extended to the cloud, and one involves workload scaling. There may be times your organization needs to ramp up a production workload beyond what the local data center can comfortably handle.
For example, consider the way insurance companies operate. For most of the year, insurance companies typically consume a predictable level of IT resources. However, there may be open enrollment periods that occur throughout the year.
These enrollment periods are especially busy for insurance companies. As a result, existing servers may not be able to handle open enrollment workloads. Rather than buy new servers to accommodate temporary spikes in demand, the companies could use public cloud.
If the company's enrollment applications are Web-based, it would be relatively easy for cloud-based Web servers to accommodate the seasonal demand. Once open enrollment is over, the cloud-based Web servers can be decommissioned.
Another advantage of cloud-based VMs is protecting businesses in case of equipment failures or physical disasters.
To protect against data center failures, some organizations build geographic clusters that span multiple data centers. Then, if a natural disaster destroys an organization's primary data centers, mission-critical workloads fail over to secondary data centers.
Building geoclusters, however, is expensive and complicated. Alternatively, organizations can use public cloud to protect mission-critical workloads using guest clusters and VM replication.
Guest clusters are clustered application servers that run on virtual hardware. Since all of the guest cluster nodes are VMs, it's possible to place some of the cluster nodes in the public cloud. That way, critical workloads can continue to function even if primary data centers go offline.
However, you probably can't include public cloud servers in existing hypervisor clusters. Public cloud providers typically don't give customers low-level access to physical servers, which makes stretching an existing hypervisor cluster to the public cloud nearly impossible.
Another way to use public cloud for business continuity is through VM replication. Not every provider or server virtualization platform supports replication, but some cloud and hypervisor combinations allow duplicate VMs to be created in the cloud and kept in sync with on-premises VMs.
Similarly, there are solutions that allow public clouds to act as relays between private data centers for VM replication. If any primary data centers go down, VM replicas can be used.
Public cloud VM deployment can be simpler than on-premises deployment because public cloud providers typically offer templates to generate new VMs. Templates are used to automate VM creation. And, because they don't require manual installation of guest operating systems, templates cut down on administrators' work. Some templates even include pre-installed applications.
Public cloud VM deployments are especially attractive for startups. Any new business incurs startup costs, but operating virtual servers in the cloud eliminates the capital expenditures associated with server hardware and software.
Cloud-hosted VMs are usually billed on a monthly basis. For a startup, this monthly charge is likely far less than the hardware and software costs required to run the workload locally. A startup organization can first run a workload in the cloud, and then later bring network load on-premises.
Managing a public cloud migration
Prepare your data center infrastructure for a cloud migration
Dig Deeper on Infrastructure (IaaS) cloud deployment strategies