By Kevin Woods, Head of Product Marketing at Scalyr

Emergence of Performance Information Platforms

Improving performance, perpetually and predictably, is what modern digital business is all about. Or at least it should be. Understanding performance, in real time, is the difference between the winners and the rest. Insight into the performance of digital systems, products, users, capacity needs, etc. is what creates competitive advantage, user satisfaction, and new product innovations.

Gathering and analyzing performance intelligence allows us to understand, measure and improve every aspect of the system, product, and business performance. Performance intelligence is about collecting performance information and making it easily available to those who can benefit from it. Many businesses today don't have a performance intelligence platform (PIP) that provides affordable, real time access to performance data, let alone one that is accessible to a high percentage of the people that can make use and sense of the data.  This means they could be under performing relative to their full potential, and it means there may be considerable upside to capture across the business.

The next year will see the emergence of performance intelligence platforms with real time insight into business, system, and user performance. The PIP is a system or platform that collects, stores, and accesses real time and historical performance data quickly and affordably, and makes it easy for people to use the data. Platforms, in the truest sense of the word, are the foundations upon which other things can be built.  This means: easy and intuitive interface, snappy results, and low cost. Anyone should be able to use the data to gain insight, just like we've seen with the rise of business intelligence platforms.

Unlike the predecessors of PIP, which siloed performance information into separate and limited databases, the PIP will be the convergence of tools and data into a single ecosystem, without today's painful tradeoffs in cost and performance.

The architectural necessity for an integrated data store has a number of benefits to users. First, users of various tools will be looking at the same underlying data to evaluate performance and solve problems. For example, if an SRE notices a user-response time problem using metrics, the engineers that have to read traces and logs to identify the problem will be using the same source log data. This single context brings consistency and fosters better collaboration between teams.

Second, using PIP allows DevOps teams to choose the right tool for the job.  SREs looking at response times may favor a different tool, even from a different vendor, than software engineering trying to debug code. Use of a common data store alleviates the need to try and buy integrated solutions from a single vendor who may, for example, have strong monitoring capabilities, but not have a strong log management offer.

Last but not least, machine learning algorithms can be better optimized when using comprehensive granular data from as many sources as possible. When users need to shorten storage durations or filter data to save costs, this may impede the ability of AIOps tools to do their job. A PIP will be the best choice for future ML/AI algorithms.

Why Now?

The costs and complexities of managing performance data today inhibits use, which means organizations are underperforming their full potential. There is nothing in the laws of physics or of business that keep these problems from being solved. 

If you face competition for customers in your industry, then the cost of doing business without a PIP is high. You may underperform across every aspect of the customer's experience, spend more money to deliver the experience than necessary, and take longer to correct problems, thus impacting customer loyalty, trust, usability and brand value. A leap ahead of the competition happens with insights into customer needs, system behaviors, capacity requirements, and trouble spots, creating new business advantages in the new year.

It is possible to ingest and cost-effectively store performance data and instantly access and search data in real time. Collecting and searching performance data has been too slow and expensive because first generation architectures are limited in scale and speed. The rise of second generation architectures provide affordable scale and provide a reason to celebrate as we roar into the 2020s.

##

About the Author