Welcome!

Microservices Expo Authors: AppDynamics Blog, Elizabeth White, Liz McMillan, Carmen Gonzalez, Anders Wallgren

Related Topics: Java IoT, Microservices Expo, Adobe Flex, IoT User Interface, Apache

Java IoT: Article

Why Averages Are Inadequate, and Percentiles Are Great

Averages are ineffective because they are too simplistic and one-dimensional

Anyone who ever monitored or analyzed an application uses or has used averages. They are simple to understand and calculate. We tend to ignore just how wrong the picture is that averages paint of the world. To emphasis the point let me give you a real-world example outside of the performance space that I read recently in a newspaper.

The article was explaining that the average salary in a certain region in Europe was 1900 Euro's (to be clear this would be quite good in that region!). However when looking closer they found out that the majority, namely 9 out of 10 people, only earned around 1000 Euros and one would earn 10.000 (I over simplified this of course, but you get the idea). If you do the math you will see that the average of this is indeed 1900, but we can all agree that this does not represent the "average" salary as we would use the word in day to day live. So now let's apply this thinking to application performance.

The Average Response Time
The average response time is by far the most commonly used metric in application performance management. We assume that this represents a "normal" transaction, however this would only be true if the response time is always the same (all transaction run at equal speed) or the response time distribution is roughly bell curved.

A Bell curve represents the "normal" distribution of response times in which the average and the median are the same. It rarely ever occurs in real applications

In a Bell Curve the average (mean) and median are the same. In other words observed performance would represent the majority (half or more than half) of the transactions.

In reality most applications have few very heavy outliers; a statistician would say that the curve has a long tail. A long tail does not imply many slow transactions, but few that are magnitudes slower than the norm.

This is a typical Response Time Distribution with few but heavy outliers - it has a long tail. The average here is dragged to the right by the long tail.

We recognize that the average no longer represents the bulk of the transactions but can be a lot higher than the median.

You can now argue that this is not a problem as long as the average doesn't look better than the median. I would disagree, but let's look at another real-world scenario experienced by many of our customers:

This is another typical Response Time Distribution. Here we have quite a few very fast transactions that drag the average to the left of the actual median

In this case a considerable percentage of transactions are very, very fast (10-20 percent), while the bulk of transactions are several times slower. The median would still tell us the true story, but the average all of a sudden looks a lot faster than most of our transactions actually are. This is very typical in search engines or when caches are involved - some transactions are very fast, but the bulk are normal. Another reason for this scenario are failed transactions, more specifically transactions that failed fast. Many real-world applications have a failure rate of 1-10 percent (due to user errors or validation errors). These failed transactions are often magnitudes faster than the real ones and consequently distorted an average.

Of course performance analysts are not stupid and regularly try to compensate with higher frequency charts (compensating by looking at smaller aggregates visually) and by taking in minimum and maximum observed response times. However we can often only do this if we know the application very well, those unfamiliar with the application might easily misinterpret the charts. Because of the depth and type of knowledge required for this, it's difficult to communicate your analysis to other people - think how many arguments between IT teams have been caused by this. And that's before we even begin to think about communicating with business stakeholders!

A better metric by far are percentiles, because they allow us to understand the distribution. But before we look at percentiles, let's take a look a key feature in every production monitoring solution: Automatic Baselining and Alerting.

Automatic Baselining and Alerting
In real-world environments, performance gets attention when it is poor and has a negative impact on the business and users. But how can we identify performance issues quickly to prevent negative effects? We cannot alert on every slow transaction, since there are always some. In addition, most operations teams have to maintain a large number of applications and are not familiar with all of them, so manually setting thresholds can be inaccurate, quite painful and time-consuming.

The industry has come up with a solution called Automatic Baselining. Baselining calculates out the "normal" performance and only alerts us when an application slows down or produces more errors than usual. Most approaches rely on averages and standard deviations.

Without going into statistical details, this approach again assumes that the response times are distributed over a bell curve:

The Standard Deviation represents 33% of all transactions with the mean as the middle. 2xStandard Deviation represents 66% and thus the majority, everything outside could be considered an outlier. However most real world scenarios are not bell curved...

Typically, transactions that are outside two times standard deviation are treated as slow and captured for analysis. An alert is raised if the average moves significantly. In a bell curve this would account for the slowest 16.5 percent (and you can of course adjust that); however; if the response time distribution does not represent a bell curve, it becomes inaccurate. We either end up with a lot of false positives (transactions that are a lot slower than the average but when looking at the curve lie within the norm) or we miss a lot of problems (false negatives). In addition if the curve is not a bell curve, then the average can differ a lot from the median; applying a standard deviation to such an average can lead to quite a different result than you would expect. To work around this problem these algorithms have many tunable variables and a lot of "hacks" for specific use cases.

Why I Love Percentiles
A percentile tells me which part of the curve I am looking at and how many transactions are represented by that metric. To visualize this look at the following chart:

This chart shows the 50th and 90th percentile along with the average of the same transaction. It shows that the average is influenced far mor heavily by the 90th, thus by outliers and not by the bulk of the transactions

The green line represents the average. As you can see it is very volatile. The other two lines represent the 50th and 90th percentile. As we can see the 50th percentile (or median) is rather stable but has a couple of jumps. These jumps represent real performance degradation for the majority (50%) of the transactions. The 90th percentile (this is the start of the "tail") is a lot more volatile, which means that the outliers slowness depends on data or user behavior. What's important here is that the average is heavily influenced (dragged) by the 90th percentile, the tail, rather than the bulk of the transactions.

If the 50th percentile (median) of a response time is 500ms that means that 50% of my transactions are either as fast or faster than 500ms. If the 90th percentile of the same transaction is at 1000ms it means that 90% are as fast or faster and only 10% are slower. The average in this case could either be lower than 500ms (on a heavy front curve), a lot higher (long tail) or somewhere in between. A percentile gives me a much better sense of my real world performance, because it shows me a slice of my response time curve.

For exactly that reason percentiles are perfect for automatic baselining. If the 50th percentile moves from 500ms to 600ms I know that 50% of my transactions suffered a 20% performance degradation. You need to react to that.

In many cases we see that the 75th or 90th percentile does not change at all in such a scenario. This means the slow transactions didn't get any slower, only the normal ones did. Depending on how long your tail is the average might not have moved at all in such a scenario.!

In other cases we see the 98th percentile degrading from 1s to 1.5 seconds while the 95th is stable at 900ms. This means that your application as a whole is stable, but a few outliers got worse, nothing to worry about immediately. Percentile-based alerts do not suffer from false positives, are a lot less volatile and don't miss any important performance degradations! Consequently a baselining approach that uses percentiles does not require a lot of tuning variables to work effectively.

The screenshot below shows the Median (50th Percentile) for a particular transaction jumping from about 50ms to about 500ms and triggering an alert as it is significantly above the calculated baseline (green line). The chart labeled "Slow Response Time" on the other hand shows the 90th percentile for the same transaction. These "outliers" also show an increase in response time but not significant enough to trigger an alert.

Here we see an automatic baselining dashboard with a violation at the 50th percentile. The violation is quite clear, at the same time the 90th percentile (right upper chart) does not violate. Because the outliers are so much slower than the bulk of the transaction an average would have been influenced by them and would not have have reacted quite as dramatically as the 50th percentile. We might have missed this clear violation!

How Can We Use Percentiles for Tuning?
Percentiles are also great for tuning, and giving your optimizations a particular goal. Let's say that something within my application is too slow in general and I need to make it faster. In this case I want to focus on bringing down the 90th percentile. This would ensure sure that the overall response time of the application goes down. In other cases I have unacceptably long outliers I want to focus on bringing down response time for transactions beyond the 98th or 99th percentile (only outliers). We see a lot of applications that have perfectly acceptable performance for the 90th percentile, with the 98th percentile being magnitudes worse.

In throughput oriented applications on the other hand I would want to make the majority of my transactions very fast, while accepting that an optimization makes a few outliers slower. I might therefore make sure that the 75th percentile goes down while trying to keep the 90th percentile stable or not getting a lot worse.

I could not make the same kind of observations with averages, minimum and maximum, but with percentiles they are very easy indeed.

Conclusion
Averages are ineffective because they are too simplistic and one-dimensional. Percentiles are a really great and easy way of understanding the real performance characteristics of your application. They also provide a great basis for automatic baselining, behavioral learning and optimizing your application with a proper focus. In short, percentiles are great!

More Stories By Michael Kopp

Michael Kopp has over 12 years of experience as an architect and developer in the Enterprise Java space. Before coming to CompuwareAPM dynaTrace he was the Chief Architect at GoldenSource, a major player in the EDM space. In 2009 he joined dynaTrace as a technology strategist in the center of excellence. He specializes application performance management in large scale production environments with special focus on virtualized and cloud environments. His current focus is how to effectively leverage BigData Solutions and how these technologies impact and change the application landscape.

Comments (1) View Comments

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.


Most Recent Comments
rtalexander 11/21/12 12:58:00 AM EST

Hey, could you post a reference or two that covers the theory and/or practicalities of the approach you describe?

Thanks!

@MicroservicesExpo Stories
Adding public cloud resources to an existing application can be a daunting process. The tools that you currently use to manage the software and hardware outside the cloud aren’t always the best tools to efficiently grow into the cloud. All of the major configuration management tools have cloud orchestration plugins that can be leveraged, but there are also cloud-native tools that can dramatically improve the efficiency of managing your application lifecycle. In his session at 18th Cloud Expo, ...
Microservices are a type of software architecture where large applications are made up of small, self-contained units working together through APIs that are not dependent on a specific language. Each service has a limited scope, concentrates on a specific task and is highly independent. This setup allows IT managers and developers to build systems in a modular way. In his book, “Building Microservices,” Sam Newman said microservices are small, focused components built to do a single thing very w...
When building large, cloud-based applications that operate at a high scale, it’s important to maintain a high availability and resilience to failures. In order to do that, you must be tolerant of failures, even in light of failures in other areas of your application. “Fly two mistakes high” is an old adage in the radio control airplane hobby. It means, fly high enough so that if you make a mistake, you can continue flying with room to still make mistakes. In his session at 18th Cloud Expo, Lee...
WebSocket is effectively a persistent and fat pipe that is compatible with a standard web infrastructure; a "TCP for the Web." If you think of WebSocket in this light, there are other more hugely interesting applications of WebSocket than just simply sending data to a browser. In his session at 18th Cloud Expo, Frank Greco, Director of Technology for Kaazing Corporation, will compare other modern web connectivity methods such as HTTP/2, HTTP Streaming, Server-Sent Events and new W3C event APIs ...
Join us at Cloud Expo | @ThingsExpo 2016 – June 7-9 at the Javits Center in New York City and November 1-3 at the Santa Clara Convention Center in Santa Clara, CA – and deliver your unique message in a way that is striking and unforgettable by taking advantage of SYS-CON's unmatched high-impact, result-driven event / media packages.
At first adopted by enterprises to consolidate physical servers, virtualization is now widely used in cloud computing to offer elasticity and scalability. On the other hand, Docker has developed a new way to handle Linux containers, inspired by version control software such as Git, which allows you to keep all development versions. In his session at 17th Cloud Expo, Dominique Rodrigues, the co-founder and CTO of Nanocloud Software, discussed how in order to also handle QEMU / KVM virtual machin...
Microservices are all the rage right now — and the industry is still learning, experimenting, and developing patterns, for successfully designing, deploying and managing Microservices in the real world. Are you considering jumping on the Microservices-wagon? Do Microservices make sense for your particular use case? What are some of the “gotchas” you should be aware of? This morning on #c9d9 we had experts from popular chat app Kik, SMB SaaS platform Yodle and hosted CI solution Semaphore sha...
How is your DevOps transformation coming along? How do you measure Agility? Reliability? Efficiency? Quality? Success?! How do you optimize your processes? This morning on #c9d9 we talked about some of the metrics that matter for the different stakeholders throughout the software delivery pipeline. Our panelists shared their best practices.
SYS-CON Events announced today that FalconStor Software® Inc., a 15-year innovator of software-defined storage solutions, will exhibit at SYS-CON's 18th International Cloud Expo®, which will take place on June 7-9, 2016, at the Javits Center in New York City, NY. FalconStor Software®, Inc. (NASDAQ: FALC) is a leading software-defined storage company offering a converged, hardware-agnostic, software-defined storage and data services platform. Its flagship solution FreeStor®, utilizes a horizonta...
Sensors and effectors of IoT are solving problems in new ways, but small businesses have been slow to join the quantified world. They’ll need information from IoT using applications as varied as the businesses themselves. In his session at @ThingsExpo, Roger Meike, Distinguished Engineer, Director of Technology Innovation at Intuit, showed how IoT manufacturers can use open standards, public APIs and custom apps to enable the Quantified Small Business. He used a Raspberry Pi to connect sensors...
Father business cycles and digital consumers are forcing enterprises to respond faster to customer needs and competitive demands. Successful integration of DevOps and Agile development will be key for business success in today’s digital economy. In his session at DevOps Summit, Pradeep Prabhu, Co-Founder & CEO of Cloudmunch, covered the critical practices that enterprises should consider to seamlessly integrate Agile and DevOps processes, barriers to implementing this in the enterprise, and pr...
The cloud promises new levels of agility and cost-savings for Big Data, data warehousing and analytics. But it’s challenging to understand all the options – from IaaS and PaaS to newer services like HaaS (Hadoop as a Service) and BDaaS (Big Data as a Service). In her session at @BigDataExpo at @ThingsExpo, Hannah Smalltree, a director at Cazena, will provide an educational overview of emerging “as-a-service” options for Big Data in the cloud. This is critical background for IT and data profes...
If we look at slow, traditional IT and jump to the conclusion that just because we found its issues intractable before, that necessarily means we will again, then it’s time for a rethink. As a matter of fact, the world of IT has changed over the last ten years or so. We’ve been experiencing unprecedented innovation across the board – innovation in technology as well as in how people organize and accomplish tasks. Let’s take a look at three differences between today’s modern, digital context...
The principles behind DevOps are not new - for decades people have been automating system administration and decreasing the time to deploy apps and perform other management tasks. However, only recently did we see the tools and the will necessary to share the benefits and power of automation with a wider circle of people. In his session at DevOps Summit, Bernard Sanders, Chief Technology Officer at CloudBolt Software, explored the latest tools including Puppet, Chef, Docker, and CMPs needed to...
The (re?)emergence of Microservices was especially prominent in this week’s news. What are they good for? do they make sense for your application? should you take the plunge? and what do Microservices mean for your DevOps and Continuous Delivery efforts? Continue reading for more on Microservices, containers, DevOps culture, and more top news from the past week. As always, stay tuned to all the news coming from@ElectricCloud on DevOps and Continuous Delivery throughout the week and retweet/favo...
In a previous article, I demonstrated how to effectively and efficiently install the Dynatrace Application Monitoring solution using Ansible. In this post, I am going to explain how to achieve the same results using Chef with our official dynatrace cookbook available on GitHub and on the Chef Supermarket. In the following hands-on tutorial, we’ll also apply what we see as good practice on working with and extending our deployment automation blueprints to suit your needs.
Let’s face it, embracing new storage technologies, capabilities and upgrading to new hardware often adds complexity and increases costs. In his session at 18th Cloud Expo, Seth Oxenhorn, Vice President of Business Development & Alliances at FalconStor, will discuss how a truly heterogeneous software-defined storage approach can add value to legacy platforms and heterogeneous environments. The result reduces complexity, significantly lowers cost, and provides IT organizations with improved effi...
Cognitive Computing is becoming the foundation for a new generation of solutions that have the potential to transform business. Unlike traditional approaches to building solutions, a cognitive computing approach allows the data to help determine the way applications are designed. This contrasts with conventional software development that begins with defining logic based on the current way a business operates. In her session at 18th Cloud Expo, Judith S. Hurwitz, President and CEO of Hurwitz & ...
Data-as-a-Service is the complete package for the transformation of raw data into meaningful data assets and the delivery of those data assets. In her session at 18th Cloud Expo, Lakshmi Randall, an industry expert, analyst and strategist, will address: What is DaaS (Data-as-a-Service)? Challenges addressed by DaaS Vendors that are enabling DaaS Architecture options for DaaS
One of the bewildering things about DevOps is integrating the massive toolchain including the dozens of new tools that seem to crop up every year. Part of DevOps is Continuous Delivery and having a complex toolchain can add additional integration and setup to your developer environment. In his session at @DevOpsSummit at 18th Cloud Expo, Miko Matsumura, Chief Marketing Officer of Gradle Inc., will discuss which tools to use in a developer stack, how to provision the toolchain to minimize onboa...