What Visibility into Machine Data Transactions Can Do for Your Organization
Machine data is about more than just anticipating outages and maintenance. When machine data is used to its fullest, it informs business decisions to drive business performance.
By Bruno Kurtic, Founding Vice President of Product and Strategy, Sumo Logic
We live in a complicated, interconnected world. With new data sources emerging across different delivery models, the number of transaction and event types is growing so quickly that it's tough to determine the root cause of application performance issues or those that otherwise impact the customer experience. This matters for two reasons. The first is obvious: a business loses money every minute an application is down or even slow, so the pressure is on to identify—and fix—issues ASAP. The second is more subtle: data on events occurring across this diverse IT architecture provides insight that can inform business strategies and drive revenue.
Done correctly, machine data is more than just an operational tool. It can actually bolster the overall trajectory of the company when it helps businesses understand event relationships and, as a result, optimize customer interactions, business processes, and security procedures.
What's Missing from Most Machine Data Analytics
It's simple: insight into transactions.
The growing number of data sources, their underlying infrastructure and the complex workflow interactions among them means that each log message these sources generate is likely not just a siloed occurrence but rather part of a series of interconnected events generated by multiple and independent engines that, in the end, are a part of one distributed transaction. Unfortunately, many of the analysis tools that IT groups use today focus on silos and can't incorporate contextual information on transactions, which limits an organization's ability to determine the root cause of issues in distributed environments.
Whether the result is critical transaction latency that impacts critical revenue-generating activities, or failure to mine insight into user and customer behavior that can help dictate business strategy and decisions, the root cause is the same: human limitations to comprehend the overwhelming amount of machine data churned out by every application, website, server, mobile device, and supporting IT infrastructure component in the enterprise. Businesses need better analytics tools.
Modern applications are designed to be highly distributed and run across private and public clouds. They are composed of decoupled micro-services that are managed separately to reduce complexity, improve uptime and agility, and enable distributed teams to both develop and operate modern applications. This is increasing the need to collect and consolidate event and other machine data from distribute systems in real-time, analyze it in such a way as to understand how transactions flow through distributed micro-services from transaction start to transaction end. An example of a highly distributed system might be a retail site that runs in a private data center but has bursting capacity that spins up and down in a public cloud. The site can be accessed by a Web browser or by a custom-built mobile application both of which have special endpoints, and uses internal and third party services for targeted promotions, shipping, and payment processing.
Making Transaction Analytics Possible
To uncover these interactions and relationships across distributed systems, transaction analytics uses algorithms to track and analyze the flow of events for each transaction of interest running through an application. Each transaction as it executes leaves behind breadcrumbs and other clues such as log lines that contain session or user ids that can shed light on the health and performance of the application. Mining the huge amounts of machine data in real time, the transaction analytics algorithms can stitch together these breadcrumbs and derive powerful insights into transaction execution at each stage such as step-by-step latencies, flow counts, errors and exceptions, and user behavior, to name a few. These insights deliver deep visibility into causal relationships between events as they occur. Transaction analytics can help reduce MTTI (mean time to identification) and expedite root-cause analysis by surfacing relationships between specific application events and improve insights into user behavior.
Transaction analytics applied to machine data decreases time associated with compiling and applying intelligence, and provides clear visualization of complex transaction relationships in real-time. These visualizations depict the flow of each transaction, translating machine data into real-time information that business and IT users can readily understand and act upon.
Transaction Analysis Helps Resolve Application Performance Issues
Transaction analysis allows companies to hone in on the root cause of issues that might impact business-critical events. For example, transaction analysis can correlate customer experience issues to provide a complete picture of how well a process works. It can also track an entire process, such as an online shopping experience, to show organizations where latency or failures occurred and map those to specific regions, devices, and other parameters.
For example, it can show how many registration attempts failed in the last 15 minutes—over the phone, over the Web, and over mobile devices—to highlight any issues with the registration process and to help reveal the root cause of the failed registrations. In another example, it can support security uncover fraudulent transactions by helping companies determine if there are there any anomalous checkout transactions and which regions they come from.
Leveraging Transaction Analysis for Insight
Transaction analysis of machine data can help drive business success by identifying user behaviors. Applying user context to output of transaction analytics enables owners of user-facing applications to understand how users behave as they interact with those applications, which populations perform which actions, at what times and frequency, how much they spend based on where they come from, and much more.
Machine data is about more than just anticipating outages and maintenance. When machine data is used to its fullest, it informs business decisions to drive business performance. Transaction analytics provides a new means of extracting intelligence from machine data in order to drive new revenue and make business decisions.
Bruno Kurtic is the founding vice president of product and strategy at Sumo Logic. You can contact the author at firstname.lastname@example.org.