Over the last few years, BYOD and the consumerization of IT have created empowered users who drive new levels of innovation and productivity. This new era has also resulted in a troubling unpredictability for IT departments. Power has shifted into the hands of IT users who bring their own devices to work and use the apps and services that they want, sometimes without the knowledge of IT.
At the same time, applications are now significantly more complex, with components distributed across multiple data centers and cloud services. What was once a nice, clean application stack of three distinct tiers (presentation layer, application server/middleware, and database) is now a rather messy mash-up of dynamically changing hardware and software configurations. And these are all connected and delivered via dozens of distinct networks (including mobile) at any given moment.
These three developments ―consumerization, application complexity, and network complexity―mean that IT no longer has much visibility nor control over the applications it’s supporting. This threatens agility, customer and partner relationships, and business productivity. In many cases, performance and user experience are not consistent for all scenarios.
To manage performance in today’s environment, CIOs must understand what the user is experiencing, what’s going on inside the app itself, and all the delivery infrastructure in between. CIOs want an end-to-end view so they can find the origin of performance issues quickly. The notion of closed-loop application performance management takes this one step further, meaning IT has the comprehensive visibility, appropriate analysis, and remediation tools to take action on that data and resolve or prevent issues.
But where do you start? Based on conversations with customers leading this charge, we’ve come up with five steps you can take:
1. Collect and filter massive volumes of data in real time.
Because of the interdependencies of applications, services, and networks, companies must monitor everything. Having contextual and “emotionless” data helps IT solve problems in a way that is not politically or culturally harmful. IT must break through the IT silos by showing the business exactly what’s happening and where. Yet this is also an enormous big data challenge. Companies need analytics tools that continuously collect granular, real-time data across the environment and quickly process and provide actionable insight. And although this is a classic big data problem, don’t assume you can solve it with a generic big data analytics package. The key to efficiency and cost effectiveness is a solution purpose-built and tuned for analyzing performance.
2. Strike the right telemetry balance. This ever-growing volume of performance event data is also prone to creating bottlenecks. If you instrument too much, with agents installed on every application and node, you run the risk of adding too much overhead to the infrastructure and slowing performance―the very situation you are trying to avoid. Instead, take a blended approach that gathers data from existing infrastructure (e.g., flow data), from purpose-built infrastructure (e.g., packet capture devices), and from agents on endpoints (e.g., server-side agents, browser agents). The real value, however, comes from tools that enable rapid filtering and analyzing of the data and then presenting key metrics back to the user.
3. Create an automated infrastructure for remediation. Understanding the problem isn’t helpful if you aren’t able to fix it. That’s where technologies such as acceleration devices, content optimization, and load balancing come into play. It’s now fairly common for companies to use WAN optimization and application delivery controller (ADC) technologies for these purposes, yet deploying them in a widespread fashion across the environment is now required. There are two requirements for remediating performance issues across the entire modern enterprise. First, you need to support all deployments options. Make sure your optimization technologies are available in hardware, software, virtual, and cloud-based form factors. Second, in order to automate provisioning and changes to these technologies, it’s critical that everything is programmable by open APIs. This allows your application performance platform to push policy changes down to any device, behind the scenes.
4. Close the performance loop. Now that you’re collecting and analyzing enterprisewide performance data, it’s time to connect that to your remediation infrastructure. With the right data delivered at the right time in concert with strong policies, you can program your optimization technologies to address specific issues. You can also integrate this with additional orchestration technologies already fueling your data center. Now you can take an application hosted in-house, for example, and move it into the public cloud for a period of time, such as for high load in Q4 or to satisfy seasonal demand. Perhaps your company has a support app for the contact center, which needs to follow the sun if you are a global business. The application workload needs to continuously move to the region experiencing prime business hours.
5. Don’t rush toward automated processes. Not all companies can and should proceed to this step. Many will find steps 1 through 4 make them considerably more agile and responsive to business needs. But if you’re ready for the last step, then it’s time to automate. Your application performance platform could perform remediation actions in a fully automated fashion: This is the nirvana state for closed loop application performance management, and it requires rock-solid policies to do without human intervention. Before you jump into full automation, consider an interim state where the app performance orchestration system provides a suggested fix for someone to manually approve. It takes time to collect data on your use cases over a few months to provide a firm understanding of the scenarios and requirements which then dictate the best policies. If you adopt closed-loop automation too quickly, you run the risk of applying an incomplete or erroneous policy. You can automate a bad outcome just as quickly as a good outcome.
Modern application performance management requires careful attention to instrumentation, efficient big data analytics processing, acceleration and delivery technologies, watertight policies, and open API support. Using a phased and measured approach, applications, networks, and devices can adapt to changing conditions automatically to preserve SLAs and avoid downtime.
Learn more in “Turn Big Data Inward with IT Analytics,” a Forrester report that helps IT infrastructure and operations leaders prepare for IT analytics that turns big data efforts inward to manage the technology services that run your business.
Riverbed, at more than $1 billion in annual revenue, is the leader in application performance infrastructure, delivering the most complete platform for the hybrid enterprise to ensure applications perform as expected, data is always available when needed, and performance issues can be proactively detected and resolved before impacting business performance. Riverbed enables hybrid enterprises to transform application performance into a competitive advantage by maximizing employee productivity and leveraging IT to create new forms of operational agility. Riverbed’s 26,000+ customers include 97% of the Fortune 100 and 98% of the Forbes Global 100.