Cloud Use Case Series, Part 2: Application Cloud

Next up in my four-part cloud use case blog series is application cloud. With this approach, N-tier applications are migrated to a cloud-friendly, stateless architecture to enable automatic increases and decreases in resource allocations based on workload and resource policies.

As I mentioned in my previous post, while application clouds actually provide the most comprehensive benefits of all the four types of use cases, some early adopters have chosen to start with smaller-scope initiatives in order to acclimate their IT and business processes and application architectures to the new model of shared computing that private cloud provides. However, the long-term objectives of most of these organizations is to eventually evolve to application cloud to get the most value from their cloud investment.

Many applications—particularly Java environments—are deployed across multiple datacenters, where each application uses its own dedicated set of resources that are static and provisioned with excess capacity in order to meet peak demand periods. These applications typically run on expensive infrastructure, operating systems and middleware (e.g. UNIX SMP, UNIX OS, Websphere/Weblogic) and are built using inflexible architectures typically associated with these tools. As a result, enterprises see only about 10-15 percent utilization of the underlying infrastructure, and incur operating costs for equipment that isn’t being used (admin time, facilities, power, hardware and software maintenance).

Some of the challenges faced by organizations before migrating to an application cloud are that CIOs and architecture teams want to reduce costs by 75 percent or more by changing from UNIX SMP servers to x86 Linux servers and application middleware, using low-cost VM alternatives by Xen or KVM to make applications more portable and flexible. At the same time, the operations team wants to become more responsive to business needs and reduce application provisioning time by 90 percent or more while increasing resource utilization from 10-15 percent to 50-60 percent or higher. This would reduce IT operational costs by 25 percent or more - all while escaping vendor lock-in and regaining control of their applications and infrastructure.

To solve these challenges, applications are migrated to a cloud-friendly, stateless architecture to enable mobility of application components. Applications are automatically deployed across a shared pool of resources using private cloud management software, such as Platform ISF, which monitors application workloads and resource usage and dynamically expands or contracts capacity according to workload and resource policies. This enables guaranteed SLA’s and ensures the most efficient use of the share infrastructure, where all activities are metered to enable application-level chargeback, so business units only pay for what they use.

One of our customers, a leading global bank, develops and manages hundred of Java applications on thousands of proprietary UNIX SMP servers. These application environments are deployed in silos, resulting in low resource utilization and high operating costs with time-consuming, manual provisioning required for deployment and modification of infrastructure.

The bank migrated applications to a shared, stateless architecture managed by Platform ISF to enable automatic expansion and contraction of capacity according to application workload and resource policies on Intel-based Linux servers.

This has enabled the bank to reduce capital costs by 75 percent and operating costs by 25 percent, with resource utilization increased to 50-60 percent. The customer can now deploy application environments within minutes rather than months, and application teams can modify and scale the infrastructure of their deployed applications without IT involvement.

Stay tuned for part 3 in my cloud use case series – test/dev cloud - next week!


Post a Comment