Cross-Platform End-to-End Messaging Transaction Analysis with OmniAnalyser White Paper © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 1 Definition of the Messaging Service Using OmniAnalyser to collect and analyze messaging-service data delivers the most complete metrics information about messages sent, received or transferred in organizations of all sizes. Such messagingservice data-collection includes all individual transactions. The delivery of a message from origin to destination is monitored end-to-end and, technically, such delivery occurs across multiple systems in the enterprise infrastructure. OmniAnalyser delivers comprehensive and business-relevant end-to-end messaging-service metrics, such as message transit times, perimeter transaction processing delays, instant messaging transaction latencies. The service metrics provide business users, as well as IT staff, with a complete and transparent view of all aspects of IT service delivery. To receive true end-to-end quality metrics, they should be “technologically agnostic”, meaning that the service descriptions and service level objectives should not depend on any particular technology such as the type of mail server or hardware involved. This makes the initial service definition understandable to the business user and allows for comparable metrics across multiple systems and when any specific technology is replaced. Furthermore, having proper end-to-end metrics facilitates proactive service management as opposed to a reactive process of responding to incidents. This can only be accomplished through careful analysis of the reasons for failures rather than by monitoring and reacting to performance alerts. Principles of End-to-End Messaging Transaction Analysis Metrics are the face of IT services turned towards the customer. Endto-end messaging transaction analysis using OmniAnalyser™ is focuses on transaction histories, where a transaction equals an email message in the e-mail case. Quality end-to-end metrics require © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 2 transaction integration across multiple platforms and systems in the organisation’s infrastructure. Starting with a transaction definition at the business level and detailing the transaction to its elements on multiple systems and platforms is a key to building business metrics. © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 3 To achieve quality metrics collection, an organization should integrate key transaction-metering across platforms and systems, no matter how large or diversified they are. OmniAnalyser™ re-assembles all data related to each messaging transaction collected from multiple servers and systems, so its database contains full information on each transmitted email. Different pieces of the same transaction are pulled together into the OmniAnalyser™ Integration Server and connected to each other based upon common attributes of each message that remain consistent across systems. Typically these are message IDs, although those IDs may not be the same for all participating systems, so multiple unique keys may have to be used for complex infrastructures. In the example illustrated by the diagrams above, London is a Sendmail server with Postfix content filtering deployed on it, whereas Grenoble is a Microsoft Exchange Server. Messages that go through both servers are counted and analysed in their entirety as seen on the Event List report. © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 4 Benefits of this Approach for the Business User End-to-end messaging transaction analysis ensures the collection of meaningful information on the quality and volume of service. Delivery delays are determined for all “real” user messages (not just for probes as in some alternative approaches). Delivery times can also be calculated for subsets of larger infrastructures, such as business units, geographies, or specific types of servers. End-to-end integration also delivers accurate volumetric data as messages are counted when passing through multiple systems. A different approach to that would deliver inaccurate numbers, which may be off by orders-of-magnitude. OmniAnalyser™ delivers auditable and accurate transaction volume data: the business user can determine the number of messages being transferred in any context within the organization, as well as in and outbound to/from the Internet. Those numbers are auditable, so they can be detailed down to each individual email and user. OmniAnalyser™ properly consolidates such information across different servers and systems, leading to correct email counts. Benefits for the Operator of the Servers There needs to be a rational, process-oriented methodology towards managing the messaging-service delivery in a transparent and efficient way. OmniAnalyser provides the ability to produce both customerfacing and operational metrics. The operational metrics are sufficiently granular for proactive problem analysis and forensics. Hypersoft OmniAnalyser™ delivers the most comprehensive reports on the technical state of the server, alerting if some threshold in analyzed data has been reached. Thanks to this function, the administrator operator of each server can always monitor the server’s technical state real-time. Another important operational benefit is the ability to precisely match the scope of the metrics to the scope of the managed systems. © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 5 Integration of End-to-End Metrics into Service Operation Scenarios Enterprise Operations needs metrics to be mapped to business units, locations, categories of users, service hours, and other parameters in the organization. All such parameters map differently to different systems in use. A perimeter application (such as antispam) may be shared across all business units, or there may be several perimeter installations in different geographies (such as Europe and North America). There may be one set of scheduled maintenance times for mobile messaging (such as Blackberry) and a different set of maintenance times for the enterprise mail system (such as Lotus Domino). Having proper metrics and, in particular, delivery times and mail traffic information is only realistic if the enterprise has a complete transaction data history in the database repository. With complete transaction information, operational metrics and subsets can be generated by filtering the queries to the repository. Otherwise it would need to re-collect raw data which is sometimes just impossible. In summary, there are two principal approaches to service metrics collection: ‐ ‐ Top-down, starting with the business service definition and proceeding to the technical metrics required to produce that; Bottom-up, looking at available technical metrics and trying to transform them into something understandable for business. With the top-down metrics collection: 9 KPIs are understood by provider and client, both internal and external 9 Clear assignment of service delivery metrics to business values 9 Standardized service portfolio 9 True end-to-end proactive monitoring for optimal service delivery In the case of messaging services, the top-down approach requires integrated transaction histories across multiple platforms. This is the only way to keep the focus on the concept of the message’s transmission (which is the key business value for a user) and not just on the “technical” health of participating servers and systems. © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 6 © Hypersoft Informationssysteme GmbH, 2008 www.re-soft.com/metrics 203 972 8462 7