Service-orientation initiatives are often driven by the need to provide multi-channel delivery of goods or converged services over the Internet. Common business applications supported by business transactions include:

• OSS/BSS
• Claims processing
• Sales lead qualification
• Online reservations

These initiatives are often executed by arranging, or "orchestrating," existing applications and infrastructure to implement business processes. A business transaction is a set of business-focused operations that involves shared components, such as SOAP services, packaged applications, EJBs, POJOs, ASP.NET, ESBs, databases and so on. However, despite the complexity of such a process, it must look like a single transaction to the business user.

Operations and Application Support staff responsible for these business systems face a number of challenges:

• Lack of end-to-end visibility across distributed transactions
• Inability to visualize and analyze business aspects of transactions
• Lack of end-to-end performance management
• Difficulty locating performance bottlenecks and the cause of errors

Neither IT nor business teams have traditionally had the visibility they need to efficiently manage composite applications or end-to-end transactions. In fact, business transactions spanning multiple systems and services have the potential to "vanish" at certain points in their flows. IT and business staff typically first discover transaction issues through cryptic log entries or calls from vexed customers. What usually follows are resource-intensive fire-drills and finger-pointing across various teams, increasing the maintenance overheads incurred by IT.

Finding a Bottleneck in a Haystack
It’s difficult for Operations to identify transaction bottlenecks since the issue could be anywhere. It could result from components in the implementation layer of the service, or might actually be located in one of many replicated services or in the infrastructure supporting the orchestrated services. Some business transactions may be long-running or asynchronous, involving human interaction and spanning multiple systems and several days. Due to this ‘weak-link’ effect, performance and capacity are key challenges for managing distributed transactions. A performance meltdown at any given point may be extremely difficult to track down, making it doubly difficult to commit to and conform to expectations for performance and availability.

The Cost is the Customer
If such unexpected conditions aren't detected proactively and diagnosed in a timely fashion, they can directly impact business outcomes. Shopping carts lose their contents, accounts fail to be provisioned, orders are lost, and packages are rendered untraceable. And while the IT team may be aware of only some components of the overall transaction, there’s certainly one person who sees everything he or she needs to—the business user. From the customer’s vantage point, the entire transaction has failed and it doesn’t matter where, how or why. Customer satisfaction is the biggest loss in this scenario.

Taking Care of Business
While it’s difficult to get a handle on the flow of messages through such systems, it can be doubly challenging for business stakeholders to find out how much business is flowing through a service, how much a single application fault costs in terms of lost sales, or how valuable one service is compared to another. But to be successful, transaction management requires awareness of the business-specific content of the message being exchanged. The business data flowing through a system is equally important as the technical information. Without business information, it is virtually impossible for business users or for IT personnel to optimize outcomes.

To prevent business transactions from turning into "black-boxes," enterprises need more granular visibility into applications and end-to-end business processes. Visibility into just one application, or only into the service layer, no longer suffices for supporting enterprise-wide business transactions.

AmberPoint offers industry’s most comprehensive solution for managing the health of composite SOA-based business applications.

Transaction Tracking

Effective business transaction management begins with transaction tracking. As with service discovery, transaction tracking provides a baseline for management by delivering visibility into the flow of processes as they propagate across the network.

The challenge, however, is to track transactions without breaking them in the process. To do so, the transaction management system must be non-invasive. That is, it cannot modify the messages in order to view transaction flows. "Tagging" messages can have a variety of dire consequences—in regulated environments, additional logging may be required as messages are modified. Worse yet, adding proprietary message "tracers" can break dependent applications. The challenge, then, is track end-to-end transactions without modifying any of the messages involved.

To address these challenges, AmberPoint non-invasively tracks transactions across a wide range of SOA interactions, including those facilitated by SOAP and XML services, messaging systems such as JMS and MQ, database calls, RMI and EJB applications, and across various runtime infrastructures such as Enterprise Service Bus (ESBs), application servers and appliances.

End-to-end Performance, Availability and Service Level Management

AmberPoint monitors system traffic in real time in order to keep enterprises appraised of the real-time behavior of each application component. It provides a rich snapshot of vital runtime data—such as throughput, availability, response times and faults—across various slices of time.

AmberPoint brings predictability, visibility and control to SOA applications by delivering comprehensive service level management for services, transactions and business processes across heterogeneous environments. AmberPoint monitors all flavors of SOA services, ranging from SOAP and XML services to virtually any application service or legacy component—such as EJBs, Plain Old Java Objects, etc. AmberPoint can instrument virtually any distributed application service.

With AmberPoint, users can define different SLAs for discrete business segments and prioritize service use by any business criteria—such as focusing on most valuable users (customers, partners, etc.) or providing the best Quality of Service during peak hours. SLAs can be set and monitored for individual services as well as composites such as processes and transactions.

Knowing the service consumers is also a critical aspect of understanding and controlling systems. AmberPoint shows specifically who is consuming the services, and organizes them into different segments for SLA-driven quality of service. Its support for detailed usage analysis and reporting over time help organizations to identify trends and revenue opportunities.

AmberPoint prevents service problems by providing early warnings, facilitating impact analysis and initiating timely response. It has the unique capability of preventing traffic spikes and overloads from impacting the system by selectively throttling traffic before peak capacity is reached. This service throttling feature is particularly valuable in protecting new SOA investments—such as SOA-enabled mainframe systems—from unexpected demand.

Transaction Diagnosis and Root-cause Analysis

AmberPoint provides sophisticated instrumentation for real-time detection, alerting, and remediation of various types of unexpected conditions—either business or technical in nature. Using AmberPoint, application support personnel can search transactions based on message content and context—such as time of arrival, message type, or client credentials—to narrow-down the root cause. They can then examine individual messages or correlated flows to understand the complete context. AmberPoint slashes the time it takes to diagnose runtime issues from hours to minutes with many other built-in tools to facilitate root-cause analysis—such as the ability to compare messages, replay transactions and submit test messages.

AmberPoint’s root-cause analysis capabilities address a range of issues common in business transactions, such stalled transactions, missing steps, faults, and application exceptions, as well as low-level issues such as incorrect data values, boundary conditions, and so on. It does this flexibly, enabling users to define the conditions of interest. AmberPoint then monitors all traffic flowing across a transaction, waiting for the moment when an exceptional event occurs.

In complex transactions, such an exercise can be difficult. Usually, application teams must cull information from widely dispersed log files. With AmberPoint, messages from all participating components within the transactions are collected and correlated, with exceptional conditions called out. Additionally, this real-time data makes it easy for Operations personnel to quickly locate the offending component, and provide application support teams with detailed information useful for debugging and repairing the problem, thus slashing the mean-time-to-repair.

AmberPoint also provides real-time action to modify the behavior of services under pre-defined conditions. For example, if the services supporting platinum customers approach thresholds at which SLAs will be breached, AmberPoint can redirect traffic via routing or load balancing with no human intervention. This provides higher levels of customer satisfaction, while preventing organizations from suffering the consequences of failed service levels.

Transaction Monitoring for Business

AmberPoint provides a rich library of portlets and sophisticated capabilities for capturing and visualizing any business metric. It does this by monitoring the content of messages in real-time, as they flow through the SOA system. Examples include key properties—such as order totals or purchase amounts—aggregated across multiple transactions.

All metrics gathering and measurement are performed non-invasively—without requiring any alteration to the application code, service logic or message traffic. This non-invasive approach enables users to customize measurements on-the-fly, without interrupting the running system or any existing applications that may be sensitive to proprietary modification of messages.