Distributed computing, falling technology prices and budget freedom have led to a proliferation of application-specific...
servers. For CIOs, this has triggered major increases in support costs from not only the multiplication of servers but their distribution throughout company locations, often away from central technology support resources.
These platforms were underused and often failed to conform to company standards, making compliance auditing and security difficult. Complications and costs created the drive for “server consolidation,” the primary stimulus for interest in server virtualization and cloud computing.
Server consolidation virtualizes a central “farm” of servers that replace a large community of disorderly, distributed servers. Cloud computing offers a similar value: replacing underused and hard-to-support local computing with a hosted service. Companies can eliminate many application-specific servers using either consolidation or cloudsourcing, so it’s essential to pick the best strategy.
When to cloudsource servers
To see whether a given server should be consolidated or cloudsourced, you need to know how much memory, CPU and storage I/O activity applications on that server generate. This information will help estimate the cost of cloudsourcing any apps and determine whether the applications use too many resources to be run in a virtual machine (VM) that shares a server with other applications. Resource-hogging applications should run on their own servers, with businesses centralizing those servers for ease of support.
Resource-hogging applications should run on their own servers, with businesses centralizing those servers for ease of support.
Data storage and exchange costs can stall or even terminate cloudsourcing projects for some companies, so it’s important to review data usage for candidate applications and factor in costs carefully before deciding. Platform as a Service (PaaS) and Software as a Service (SaaS) can reduce maintenance costs and license fees for an application or platform software. The best applications for the cloud have limited data usage and do not access gigantic corporate data stores.
SaaS, IaaS or PaaS: Choosing a cloud service model
After vetting your apps for the cloud, the next step is to determine what cloud service model will best suit the applications. Many users make a fundamental mistake of trying to duplicate the virtualization-based server consolidation architecture in the cloud using Infrastructure as Service (IaaS). While this approach nearly always works because IaaS is the most general and flexible cloud service available, it may not be the most cost-effective choice. To determine if the cloud option is a good alternative to server consolidation, ask these three questions:
- Does the application’s resource use fit with the cloud service pricing model? Any app that accesses a massive database will incur high storage charges; an application that runs 24/7 will incur a higher usage charge. Look for low resource use, particularly in terms of data storage.
- Does the cloud service allow a business to pick a single cloud provider to support many applications? Supporting multiple cloud providers is costly; it’s technically more complex to hybridize multiple providers with applications that remain in the corporate data center. Find a service that offers the best cost versus support ratio for the largest range of candidates.
- Does the cloud service replace as much of the total application platform -- hardware, operating system (OS), middleware and application software -- as possible? Support for user components is your responsibility. This includes software licenses, maintenance and updates, installation and technical support. Higher-layer cloud offerings, such as PaaS and SaaS, displace more cost, which can make them more valuable, particularly for small or remote businesses with limited technical support resources.
Most companies should determine first whether an application can be replaced by SaaS, either from the same company or from one with the same mission. For example, companies that host customer relationship management (CRM) on independent servers often find that Salesforce.com is a cheaper option than consolidating CRM servers into a single data center farm using virtualization.
Higher-layer cloud offerings, such as PaaS and SaaS, displace more cost, which can make them more valuable, particularly for small or remote businesses with limited technical support resources.
Deciding whether to use IaaS or PaaS to support a given application requires more analysis. PaaS include OSes and middleware, so cloud service costs usually cover software and support. Savings and support can be major deciding factors, particularly if a corporate license doesn’t already cover an application’s platform services. No license may also mean that there is no in-house support. Cloudsourcing could eliminate the need to hire or train support staff. PaaS installments typically require less cloud management and are easier to hybridize with in-house application services that use the same platform.
On the other hand, IaaS is a more general approach and the broadest option available. If you don’t want multiple services and providers, IaaS cloud services are a better alternative to consolidation than PaaS for a very diverse mix of applications.
The cloud won’t be a good alternative to consolidation for every application. Many companies will have some in-house IT applications -- even if they use the cloud. Hybridization of public cloud and private data centers will become a vital component. Having a single cloud model and provider will ease a transition to hybrid clouds and reduce support costs as applications -- in the cloud and in the data center -- evolve. Likely, the simplest model for your company to manage will be the best.
ABOUT THE AUTHOR
Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982.