7.3 Managing Service Level Agreements

Building a management strategy to deliver high-quality storage services is an increasingly strategic requirement for enterprises . In order to provide an effective and meaningful level of service, organizations need to set appropriate business objectives; design, test, and set standards for equipment; establish processes that support service levels; and build tools to automate, measure, and support the process.

Meaningful service delivery is now the mantra of today's IT organizations. To build crossfunctional agility up and down the organization, excuses are no longer acceptable regarding downtime, performance, and resource allocations . Many IT organizations want to offer SLAs to their end users as a way to illustrate service and commitment. In addition, the SLA measures the performance of the IT staff by how well it provides a valued service. This generates a risk/reward profile that fosters service development, predictable cost reductions, and an improved end- user experience. But an SLA is more than a piece of paper; it requires an organizational commitment wrapped in people, process, and systems that make the SLA tangible .

While it may be easy to understand the business value of service levels, supporting SLAs create operational challenges that need to be addressed. Several supporting management processes lie at the foundation of building SLA-ready storage environments. These management processes are designed and executed in an effort to build and deliver dependable storage environments and infrastructures that can have service levels written against them.

In this section we explore the details and fundamentals for delivering meaningful SLAs:

1. Disciplined adherence to business objectives

2. Process documentation

3. Infrastructure certification

4. Standardization and security

5. Centralization of key management tasks

6. Effective management software that normalizes the entire environment

7.3.1 Discipline and Business Objectives

One of the first activities associated with delivering an SLA-ready storage environment is defining measurable business objectives, measured through real-time metrics such as reliability and availability. The objectives are delivered through a combination of discipline and process that cuts across all organizational activities and constitutes an effective storage management foundation.

Organizational discipline is a difficult metric to track and measure. Discipline is one of those "soft" skills that can easily be lost if the day-to-day work of a company does not reinforce the activity's importance. The value of people following processes in a repetitive and predictable fashion (organizational discipline) creates significant efficiencies and reliability within the storage environment. Discipline consequently directly supports the SLA. Without a disciplined organization, the activities associated with engineering, command, and execution of the SLA become disjointed and less meaningful. For example, it is almost impossible to effectively monitor the storage infrastructure if the organization has not taken the time and effort to create and design centralized configurations. Thus, the discipline to create centralized designs enables effective monitoring.

Relationships exist between disciplined design and ongoing processes for superior management. For instance, to fully support the SLA business objectives, an enterprise must institute effective change management within its infrastructure. In fact, research by the Gartner Group states that enterprises can never achieve more than 99 percent system availability without effective change-control processes. ^[1] By combining optimal storage architecture design and effective ongoing management processes, organizations can potentially realize higher availability levels within their infrastructures.

^[1] Donna Scott, Surviving in a 24/7 World , Gartner, October 2001, p. 16.

The storage SLA process supports effective delivery and sets the tone for the entire chain of events revolving around storage management. As detailed in Figure 7-6, there are three phases in this process: plan, execute, and manage.

Figure 7-6. The storage SLA process.

The plan phase determines the promised items and whether they can be supported by the infrastructure deployed. The planning phase creates the SLA, and the subsequent phases support and execute on the commitment. The execute phase comprises a series of checkpoints to ensure the right infrastructure is in place. It culminates in the activation of the service and represents the beginning of ongoing service delivery. The manage phase contains command and control activities and incorporates critical business processes such as billing and customer satisfaction and support.

Creating business objectives around the SLA creates an important link between business goals and the activities required to deliver highly available and scalable storage environments. It also reinforces the discipline required for flawless execution in support of the SLA.

7.3.2 Process Documentation

While we have already discussed the vital importance of effective processes, it is critical to ensure the documentation of the processes. Documented procedures remove ad-hoc reactions from the equation. Giving those responsible for managing the data storage infrastructure access to the set of predefined problem-solving frameworks accomplishes this task. If these frameworks are followed according to their original intent, the operator can depend on predictable reactions from all parts of the infrastructure. Predictable reactions from the infrastructure allow the operator to manage contentions within the environment, aiding in maintaining the SLA. The documentation must be readily available and accessible by anyone who requires it. Without accessibility, the process and documentation are virtually worthless.

7.3.3 Certification

Certification provides organizations with a clear set of infrastructure choices to support various classes of SLAs. This requires filtering the vast stream of available vendor products and then funneling a select few into manageable, fully understood , carefully engineered designs that meet the storage needs of any application. Creating a lab environment and testing procedures to determine the suitability of available products accomplish this goal. Findings in the lab become blueprints for installing and operating proven storage solutions. This ensures that installations around the world in Tokyo, London, and New York are standardgiving all end users consistent, guaranteed service levels.

The certification process should review potential technology to see how it performs in each of the following 10 categories. It is important to note that this is not the testing phase, but rather an initial survey to see whether a component should be considered for implementation within the infrastructure.

Certification Performance Categories

1. Scalability

2. Manageability

3. Host support

4. Features

5. Cost

6. Reliability

7. Performance

8. Security

9. Vendor support

10. Rapid deployment

If the technology passes these steps and is deemed worthy of inclusion within the infrastructure, then the certification phase turns to the testing process. Lab testing should follow a process in which the end result is an SLA-supportable standard design.

A standard infrastructure testing process should include the following:

• Step 1 : Review the target requirements and create test plan.

• Step 2 : Run tests and publish results.

• Step 3 : Create and define the standard implementation design.

• Step 4 : Document design considerations.

• Step 5 : Implement pilot deployments.

7.3.4 Standardization

Standardization helps support the management tasks of the SLA more efficiently , while minimizing risks. The effort to standardize has ripple effects across many components , including

• Process documentation : The cornerstone checkpoint to ensure that the infrastructure and operating procedures are standardized.

• Certification : The certification process researches and generates standard infrastructure configurations.

• Management software : Depends 100 percent on the standard deployment of the storage infrastructure to operate correctly.

• Centralized operations : Relies on a set of operating standards to ensure proper resolution to change, event, and problem management issues.

• Security : The assessment of all security considerations related to onsite and offsite data management.

• Monitoring : Dependent on standardized infrastructure deployment and operating standards that guarantee the health of the environment.

Standardization reinforces operational agility. Effective standardization compartmentalizes the infrastructure and associated tasks in an orderly way that allows the IT staff to divide the labor into segments. These segments contain critical path issues and promote problem-solving ownership from within the staff, creating a more agile and effective delivery team.

7.3.5 Centralization

Centralization for storage management provides superior command and control with the decision-making pieces in one place. In addition, centralization makes command and control highly scalable and flexible as the storage under management grows and changes over time.

Centralization also helps to reinforce installed processes and promote discipline. It is easier to train and buttress process skills and simpler to build accountability and incentive metrics within the operations staff. Figure 7-7 illustrates and identifies the interrelationships between the phases and roles within the centralized framework.

Quality Assurance : The initial configuration and service level should be checked for quality assurance during the activation process. This is where the end-user architecture is introduced into the infrastructure. The quality assurance role is a four-stage process, which ensures that

1. Appropriate contacts are known.

2. The network operation center (NOC) personnel can monitor and receive "traps" from the architecture.

3. The end-user network is redundant and secure.

4. The information is documented and understood.

Asset Tracking : An asset repository should track the physical configuration and topology information as well as contacts, documentation, and the history of changes made to an environment. Through software integration, this information should be automatically linked to the change and event management processes and software.

Change Management : Centralization allows changes to be documented, tracked, timed, and approved by a unified quality assurance team. Once changes are complete, the information can be captured and integrated into a central asset-tracking repository.

Event Management : The implementation of event management software should trigger "traps" for any event. Event management combined with the asset-tracking system and trained technicians enables the end user to intelligently react to any event detected within the environment. Through integration with the asset-tracking system, NOC operators will have immediate access to all troubleshooting documentation.

Problem Management : Once SLA- impacting events are detected, NOC operators should follow the escalation procedures outlined in the SLA, follow detailed processes to initiate event tickets, and own the problem until resolution. NOC operators should act as bandleaders to ensure all aspects of the SLA are well coordinated across the infrastructure and organization.

Figure 7-7. Description of the relationships between centralization phases and roles.

Centralization is the glue that holds all the moving pieces within the infrastructure together. Without a well-coordinated and centralized effort, the SLA and therefore the operational benefits are at risk.

7.3.6 Normalizing Management Software

As discussed earlier in this chapter, software that acts like an underlying operating system for the storage infrastructure helps deliver storage SLAs. This is the software layer that normalizes the storage infrastructure (see Figure 7-3) by bringing intelligence to the core of the network. Without this piece, intelligence will be confined to the individual parts that make up the edges of the network, preventing a full view across the multiple platforms and limiting the scalability and ultimately the manageability of the infrastructure.

Specific value created from implementing this type of management framework includes

• Accurate and consistent information across a multivendor, distributed infrastructure

• Seamless integration with storage and data management applications, as well as with end-user applications to form a scalable storage and data management environment

• Standardization and automation of best practices for supporting the SLA

• More cost-effective , SLA-ready storage environment

An operating system is the fundamental software that allows applications to run on specific hardware platforms. Programs, such as Microsoft Word and PowerPoint, communicate through an application programming interface (API), which precludes the need for users to even think about the existence of an operating system. These interfaces provide access to the operating system, which in turn controls the machine.

An effective storage management software package should be the interface that frees end users from having to focus on the hardware. The software should act as the operating system that makes this happen by providing services to the management applications with which it is integrated. Through a hardware integration layer similar to that of traditional operating systems, the storage operating system should manage the resources of, execute changes to, and track the status of the entire infrastructure.

Managing storage SLAs should be treated with the appropriate attention by any IT organization. The benefits clearly drive towards creating more businesswide agility by removing the downside risk and managing application expectations. But the cost and complexity of designing and implementing such a strategy should not be underestimated. As various vendors continue to develop storage management software, the task of creating SLA-ready storage infrastructures becomes more simple and straightforward to implement. But to date, no silver bullet exists. This leaves the burden of accomplishment squarely on the shoulders of the IT and development staff to build a framework that creates meaningful business benefit without constraining the core needs of the entire organization.