Best Practice Priniciples - Monitoring

The main objectives of monitoring and measuring your ADI are to ensure responsiveness, confidentiality, integrity, and availability of information systems. The benefits of monitoring and measuring components of the ADI include the ability to:

• Provide visibility into all layers.  An ADI approach allows the connection from the user to the application to be monitored as a whole, instead of on a component-by-component basis.
• Dynamically provision and tune the ADI. Monitoring provides the information needed to adjust the usage of the applications to meet the needs of the end-user and the business.
• Prove the validity of events.  This capability is used for forensic analysis, and to prove compliance, provide a degree of non-repudiation, and perform business process analysis.
• Identify points of failure and provide seamless recovery.  Monitoring and measuring technologies look for single points of failure, and identify application failures with their symptoms and root cause. By using this functionality, the system proactively and autonomically identifies failures and reroutes the delivery accordingly.
• Measure application performance as perceived by the end-user, and identify opportunities for optimization. Measuring technologies look for latency, low bandwidth, packet loss, business process delays, workflow halts, and general inefficient behaviors.
• Measure against SLAs.  Monitoring functionality can map to SLA system, and provide insight into SLAviolations and trends.
• Participate in capacity planning. Measuring functionality helps gauge the appropriate capacity for right-sizing the system, identify green computing opportunities, and perform resource leveling.
  

The criteria to consider when designing ADI monitoring and measurement should always include:

• Regulatory compliance.  Compliance regulations require a level of monitoring and measurement to prove validity of the information provided.
• Service level agreements.  SLAs specify reporting and minimum service levels that must be monitored and reported.
• Monitoring functionality.  A dynamic coupled infrastructure cannot exist without a fully integrated monitoring element, as the use of monitoring is inherent to a policy-driven autonomic ADI.

Operating automation.  If an event is not resolved or performance becomes degraded without manual intervention, a form of automatic response should assume control. ADI should be considered an organic system that should recover and act based upon pre-determined policy elements, to achieve application delivery objectives.

There can also be criteria against monitoring certain parts of the infrastructure, such as:

• Privacy/legal issues (based on region/country).  Certain countries and regions do not allow sensitive information (for example, identifiable personal information) to be recorded, reported, or aggregated.
• Security for data sensitivity and privacy. There are various laws stating that sensitive information, such as passwords and credit card numbers, must not be stored in logs.
• Duplication of effort/data.  If monitoring is already being performed sufficiently by another process, the data collected by the original monitor should be used, instead of a second process trying to collect the same data.

The guiding principles for application delivery monitoring and measurement are:

• Applications should continue to function, regardless of what the user does purposefully or inadvertently. The monitoring function must not be intrusive to the end-user or influenced (outside of the information recorded) based on user action.
• Applications should continue to be delivered to users, regardless of service disruptions within the infrastructure. The infrastructure should be able to detect faults and seamlessly re-route connections to the most appropriate ADI component (for example, the WAN application delivery controller, hypervisor, or end-user device).
• Monitoring must be performed outside of the normal stream of operations, so that the ADI operation is observed, and not degraded. Monitoring should also be tamper-resistant, where possible, with a minimum functionality of tamper-evident logging and reporting.
• Measuring should be performed from the end-user-experience perspective, as well as the system monitoring perspective. The monitoring function should consider the totality of the end-user experience in addition to aggregating and reporting on individual event trails for specific components of the ADI. Monitor and measure from different perspectives and different points within ADI, so that the individual nature of the ADI elements can be properly gauged.
• Monitoring should be performed end-to-end, to provide a holistic view that allows the administrator to view aggregated performance, as well as individual component performance. Results should be able to be viewed from a workflow perspective (business seasonality for usage, no usage, or bad usage). ADI provides for a workflow as the end-user interacts with applications and business processes. Increasingly, business-level language and policy must be monitored within the ADI. This includes business seasonality as well as what aspects of the application are not being used (for example, accounts or features that have not been active in X number of days), what aspects are being attacked (for example, SQL injection attempted against a login page), and what aspects are overused (for example, every call to the home page of a Web application results in multiple sessions, because of poor linking and redirection).
• Measurement should also be able to detect external factors that are influencing application performance; for example, slow load of page graphics from a third-party Web site in a Web application.