Over the past two years, we have seen business transaction management (BTM) technologies make significant progress in terms of deployment and data gathering techniques. “Traditional” BTM solutions that are expensive to deploy, tough to scale and time consuming to maintain are losing ground to alternative “lightweight” approaches that have made it much easier for the mainstream to adopt this critical dimension of application performance management (APM). Techniques to capture granular transaction data at the network level have given IT operations and managed services teams access to the grass roots information of every business transaction, without requiring agents, transaction tagging or code changes.
These technological advancements, coupled with the increasing complexities of managing multi-tiered applications environments, have resulted in BTM becoming the fastest growing segment of application performance management (APM) tools today.
But a BTM tool is still only as good as the intelligence it provides. And converting high volumes of transactional performance data into knowledge that IT teams and business folks can efficiently access, analyze and act on is challenging to do.
Will Capelli of Gartner Research identifies five key dimensions that define the value of APM going forward – end-user experience monitoring, runtime application architecture discovery, user-defined transaction profiling, component deep-dive monitoring, and analytics. All of these have implications for BTM tools:
Runtime architecture discovery and mapping
Increasing complexity, particularly in the Cloud, is driving interest in discovery and mapping network topology. The push is on to define automated IT service dependency mapping that can support multiple management objectives, including the mapping of critical transaction paths through network architectures comprised of new complexities like software-as-a-service, mobile, virtualization or cloud-based services.
User-defined transaction profiling
Monitored transactions need to be profiled with respect to what an end-user views as a transaction. Transactions at the system level are important – but only in the context of applications driven by users. Transactions should relate directly to end-user activity and business processes.
End-user performance monitoring
APM has long included end-user experience monitoring, particularly on the Web. However, it needs to be extended to a greater variety of application interfaces, application types, and interaction modalities. In addition, greater detail and sophistication is needed to extend user experience beyond simple task duration and page waits.
IT management convergence
As IT management tools evolve and converge, their requirements will generate redundant needs and demands. Technologies such as BTM will need to tie into common repositories, databases, and services that are used by other management solutions. For example, while existing CMS and CMDB systems contain critical information needed by other systems, they will be strained by the demands of real-time performance management.
The Unified Transaction Model (UTM) was created to bridge the granular transaction intelligence gathered by BTM tools into all of the key dimensions mentioned above. UTM takes performance analysis to a new level by providing simultaneous real-time transaction assembly across multiple network links, for multiple application protocols, supporting multiple applications.
The UTM provides the framework IT operations and business teams need to couple transactions to a service topology derived from runtime architecture discovery. It relates low-level transactions to the transactions of end-users at interface services. It provides metrics for each of its levels, including those specific to the end-user experience. And it supports stakeholders at each level of the IT hierarchy, from network to business, to share and relate common performance management views of a complex application system.
The UTM defines four levels:
- Network links
- IT services
- End-user applications
- Business processes
Transaction types are defined for each level. Link transactions correspond to the exchanges between network interfaces governed by protocols like TCP/IP. Service transactions are composed of exchanges of application messages between running processes. Application transactions are specific to the actions of an end-user at an interface service and run end-to-end across the system. Business transactions reflect the tasks that compose a specific business process.
Each type of transaction is composed of transactions from the preceding level – for example, a certain set of network packets might define a link transaction specific to the transmission of an application message, several of which may be correlated together to form a service transaction. One or more service transactions across several different links might compose an end-user application transaction, and one or more application transactions form a business transaction that accomplishes a specific task.
UTM not only defines an explicit relationship between events at each level, it also measures and characterizes the performance with targeted metrics. Network-level measurements quantify the behavior of link transactions at the bottom of the hierarchy while business process metrics are applied to business transactions at the top. The performance of each layer can be directly compared and contrasted with every other layer. Now the IT infrastructure can be managed in terms of the business that it was intended to support.
Rapid changes in the management technology landscape demand exceptional responses from performance monitoring vendors. The UTM answers the call for BTM, by providing the unifying framework and data structure IT teams and business folks need to simplify analysis of transaction performance in the context of end-user experience, multiple application components, and network topologies that have become more modular, redundant, distributed and dynamic.