IT service management

Cloud computing solutions are implemented using service management frameworks. In designing any solution, the cloud solution architect must account for how well the organization is prepared to manage, operate, and continually improve that service. The most widely adopted industry standard process for efficiently and effectively providing IT service management (ITSM) is based on the Information Technology Infrastructure Library (ITIL). ITIL structures ITSM into four domains:

• IT infrastructure: The technology components directly related to an IT service, for example, a Red Hat Enterprise Linux (RHEL) OS instance running on a server.
• Supporting services: The underlying infrastructure required to operate the customer-facing IT services, for example, the DNS server required to reach the RHEL instance by use of the hostname. Supporting services could be referred to as IT-internal services.
• IT service: The services requested by customers. Each IT service is implemented using the corresponding IT infrastructure. The IT service, therefore, complements a set of IT infrastructure components by adding service definitions such as SLA information and cost. In our example, the RHEL instance could be offered as a gold and silver service. Gold might include 24 x 7 support while silver offers 8 x 5 support at a lower cost.
• ITSM framework: Standards and processes that orchestrate all the activities required to deploy an IT service. This is not a collection of infrastructure components but rather the framework used for the deployment, operation, and decommissioning of IT services. The ITSM framework ties the IT infrastructure, supporting services and IT services together and provides the needed operational functionality. IT services are presented to potential customers, they, in turn, need to be able to order them:

Figure 5:  IT service management framework

The ITSM framework should not be confused with the deployment of an IT infrastructure required by a service. The focus is on orchestrating all the activities required to deploy an IT service as opposed to a collection of infrastructure components. If the customer requests the RHEL gold service, it is not only about deploying the OS image using Red Hat Satellite, but also about modeling the service in the Configuration Management Database (CMDB), configuring the event and impact management, making sure that the corresponding OLA are rolled-up into the promised SLA, and reporting it in a service view on a customer-facing portal. The ITSM framework ties everything together into a coherent service offering.

In addition to implementing ITSM, the following best practices should also be practiced:

• Enforcement of brutal standardization across the enterprise: A small number of non-optional constraints is often the most effective means of achieving agile governance. Jeff Bezos at Amazon famously mandated in 2002 that "All [Amazon] teams will henceforth expose their data and functionality through service interfaces" that could eventually be exposed to a public-facing market. The form and style of the interfaces were left to the teams to determine, but critically, anyone who didn't follow the edict was subject to termination. Vigorous enforcement of a lightweight set of requirements is a recurring theme in successful modern IT management. If this is not possible to globally enforce due to organizational considerations, it becomes even more important to demonstrate success via well-scoped pilot projects that can showcase the new model.
• IT standardization: Standardization is critical to gaining operational efficiency, reducing overall cost, and reducing the time required to deliver new capabilities. Across all commercial industries, average standardization savings are on the order of a 2/3 reduction of servers. Increasing the utilization percentage realizes these. Most of the economic value in this type of transition is accomplished through standardization in the software development platform and support personnel efficiencies enabled by standardized operational processes, as shown:

Figure 6: Achieving a shared services environment

• IT change management standardization (processes and tools): As part of the adoption of ITIL, it is critical that the organization has positive knowledge of its IT assets. This is typically the first transformational step towards ITIL. Once a baseline of assets is collected, configuration control and processes must be put in place and religiously followed. This is needed to efficiently handle incident management (restore services, analyze incident types and trends, and improve communications), problem management (root cause analysis, document known errors), and change management (reduce unexpected outages, track approvals for compliance, new service support). Variations in these processes are problematic and are easily seen by the customer. The goal is to provide transparent, efficient, and consistent processes and services.
• Transition from customer-mandated to a customer-focused model: In a customer-mandated model, the solution design is primarily driven by a targeted end user's requirements and dictates. This approach can lead to wide technical variations across operationally similar solutions. In a customer-focused environment, the architects design solutions that can be used to meet a broad marketplace of users. These offerings are focused on the target audience for the provided services. For example, Amazon Web Services provides only limited patched versions of windows and Linux/Unix. This allows them to provide these service solutions at a very inexpensive rate on-demand to customers.
• Leverage all applicable shared services: The interoperability and efficiencies gained through cloud-based offerings are depicted in the following diagram. As more and more components become standardized the level of effort to provide them decreases as well as the overall system complexity. As the enterprise and the cloud offering matures, the enterprise performs more and more of the work performed and eventually, the business process-specific applications can be deployed as specialized IT service extensions.
• Use common standardized delivery patterns: Ideally, all cloud computing solutions should be designed as real-time aggregations of existing cloud services. The following diagram depicts the concept of providing a common set of standardized delivery patterns. In this diagram, the far left depicts some enterprise-wide patterns or sets of software that could be used to support specific lines of business. The diagram also includes the notion of a common development environment. This common development environment can be a PaaS or a standardized company environment that uses specified software components for targeting specific deployment patterns. By matching a certified and accredited development environment to a specific certified and accredited operational environment, an organization can rapidly develop and deploy industry-accredited solutions quickly.

Figure 8: Common standardized delivery patterns