Welcome!

AJAX & REA Authors: Liz McMillan, Plutora Blog, AppDynamics Blog, Elizabeth White, Jnan Dash

Related Topics: Java, SOA & WOA, Adobe Flex, AJAX & REA, Apache

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We recognize that the average no longer represents the bulk of the transactions but can be a lot higher than the median.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

@CloudExpo Stories
"At Harbinger we do products as well as services. Our services are with helping companies move their products to the cloud operating systems. Some of the challenges we have seen as far as cloud adoption goes are in the cloud security space," noted Shrikant Pattathil, Executive Vice President at Harbinger Systems, in this SYS-CON.tv interview at Cloud Expo, held Nov 4–6, 2014, at the Santa Clara Convention Center in Santa Clara, CA.
After a couple of false starts, cloud-based desktop solutions are picking up steam, driven by trends such as BYOD and pervasive high-speed connectivity. In his session at 15th Cloud Expo, Seth Bostock, CEO of IndependenceIT, cut through the hype and the acronyms, and discussed the emergence of full-featured cloud workspaces that do for the desktop what cloud infrastructure did for the server. He also discussed VDI vs DaaS, implementation strategies and evaluation criteria.
Things are being built upon cloud foundations to transform organizations. This CEO Power Panel at 15th Cloud Expo, moderated by Roger Strukhoff, Cloud Expo and @ThingsExpo conference chair, addressed the big issues involving these technologies and, more important, the results they will achieve. Rodney Rogers, chairman and CEO of Virtustream; Brendan O'Brien, co-founder of Aria Systems, Bart Copeland, president and CEO of ActiveState Software; Jim Cowie, chief scientist at Dyn; Dave Wagstaff, VP ...
“The year of the cloud – we have no idea when it's really happening but we think it's happening now. For those technology providers like Zentera that are helping enterprises move to the cloud - it's been fun to watch," noted Mike Loftus, VP Product Management and Marketing at Zentera Systems, in this SYS-CON.tv interview at Cloud Expo, held Nov 4–6, 2014, at the Santa Clara Convention Center in Santa Clara, CA.
Bring the world's best IaaS to the world's best PaaS. In their session at 15th Cloud Expo, Animesh Singh, a Senior Cloud Architect and Strategist for IBM Cloud Labs, and Jason Anderson, a Cloud Architect for IBM Cloud Labs, shared their experiences running Cloud Foundry on OpenStack. They focused on how Cloud Foundry and OpenStack complement each other, how they technically integrate using cloud provider interface (CPI), how we could automate an OpenStack setup for Cloud Foundry deployments, an...
"Application monitoring and intelligence can smooth the path in a DevOps environment. In a DevOps environment you see constant change. If you are trying to monitor things in a constantly changing environment, you're going to spend a lot of your job fixing your monitoring," explained Todd Rader, Solutions Architect at AppDynamics, in this SYS-CON.tv interview at DevOps Summit, held Nov 4–6, 2014, at the Santa Clara Convention Center in Santa Clara, CA.
"Desktop as a Service is emerging as a very big trend. One of the big influencers of this – for Esri – is that we have a large user base that uses virtualization and they are looking at Desktop as a Service right now," explained John Meza, Product Engineer at Esri, in this SYS-CON.tv interview at Cloud Expo, held Nov 4–6, 2014, at the Santa Clara Convention Center in Santa Clara, CA.
The consumption economy is here and so are cloud applications and solutions that offer more than subscription and flat fee models and at the same time are available on a pure consumption model, which not only reduces IT spend but also lowers infrastructure costs, and offers ease of use and availability. In their session at 15th Cloud Expo, Ermanno Bonifazi, CEO & Founder of Solgenia, and Ian Khan, Global Strategic Positioning & Brand Manager at Solgenia, discussed this shifting dynamic with an ...
More and more file-based and machine generated data is being created every day causing exponential data and content growth, and creating a management nightmare for IT managers. What data centers really need to cope with this growth is a purpose-built tiered archive appliance that enables users to establish a single storage target for all of their applications - an appliance that will intelligently place and move data to and between storage tiers based on user-defined policies. In her session a...
Recurring revenue models are great for driving new business in every market sector, but they are complex and need to be effectively managed to maximize profits. How you handle the range of options for pricing, co-terming and proration will ultimately determine the fate of your bottom line. In his session at 15th Cloud Expo, Brendan O'Brien, Co-founder at Aria Systems, session examined: How time impacts recurring revenue How to effectively handle customer plan changes The range of pricing a...
Mobile, the cloud and data have upended traditional ways of doing business. Agile, continuous delivery and DevOps have stepped in to hasten product development, but one crucial process still hasn't caught up. Continuous content delivery is the missing limb of the success ecosystem. Currently workers spend countless, non-value add hours working in independent silos, hunting for versions, manually pushing documents between platforms, all while trying to manage the continuous update and flow of mul...
Over the last few years the healthcare ecosystem has revolved around innovations in Electronic Health Record (HER) based systems. This evolution has helped us achieve much desired interoperability. Now the focus is shifting to other equally important aspects - scalability and performance. While applying cloud computing environments to the EHR systems, a special consideration needs to be given to the cloud enablement of Veterans Health Information Systems and Technology Architecture (VistA), i.e....
The Industrial Internet revolution is now underway, enabled by connected machines and billions of devices that communicate and collaborate. The massive amounts of Big Data requiring real-time analysis is flooding legacy IT systems and giving way to cloud environments that can handle the unpredictable workloads. Yet many barriers remain until we can fully realize the opportunities and benefits from the convergence of machines and devices with Big Data and the cloud, including interoperability, ...
IBM has announced a new strategic technology services agreement with Anthem, Inc., a health benefits company in the U.S. IBM has been selected to provide operational services for Anthem's mainframe and data center server and storage infrastructure for the next five years. Among the benefits of the relationship, Anthem has the ability to leverage IBM Cloud solutions that will help increase the ease, availability and speed of adding infrastructure to support new business requirements.
Vormetric on Wednesday announced the results of its 2015 Insider Threat Report (ITR), conducted online on their behalf by Harris Poll and in conjunction with analyst firm Ovum in fall 2014 among 818 IT decision makers in various countries, including 408 in the United States. The report details striking findings around how U.S. and international enterprises perceive security threats, the types of employees considered most dangerous, environments at the greatest risk for data loss and the steps or...
Companies today struggle to manage the types and volume of data their customers and employees generate and use every day. With billions of requests daily, operational consistency can be elusive. In his session at Big Data Expo, Dave McCrory, CTO at Basho Technologies, will explore how a distributed systems solution, such as NoSQL, can give organizations the consistency and availability necessary to succeed with on-demand data, offering high availability at massive scale.
Code Halos - aka "digital fingerprints" - are the key organizing principle to understand a) how dumb things become smart and b) how to monetize this dynamic. In his session at @ThingsExpo, Robert Brown, AVP, Center for the Future of Work at Cognizant Technology Solutions, outlined research, analysis and recommendations from his recently published book on this phenomena on the way leading edge organizations like GE and Disney are unlocking the Internet of Things opportunity and what steps your o...
The move in recent years to cloud computing services and architectures has added significant pace to the application development and deployment environment. When enterprise IT can spin up large computing instances in just minutes, developers can also design and deploy in small time frames that were unimaginable a few years ago. The consequent move toward lean, agile, and fast development leads to the need for the development and operations sides to work very closely together. Thus, DevOps become...
In their session at @ThingsExpo, Shyam Varan Nath, Principal Architect at GE, and Ibrahim Gokcen, who leads GE's advanced IoT analytics, focused on the Internet of Things / Industrial Internet and how to make it operational for business end-users. Learn about the challenges posed by machine and sensor data and how to marry it with enterprise data. They also discussed the tips and tricks to provide the Industrial Internet as an end-user consumable service using Big Data Analytics and Industrial C...
SYS-CON Media announced that Splunk, a provider of the leading software platform for real-time Operational Intelligence, has launched an ad campaign on Big Data Journal. Splunk software and cloud services enable organizations to search, monitor, analyze and visualize machine-generated big data coming from websites, applications, servers, networks, sensors and mobile devices. The ads focus on delivering ROI - how improved uptime delivered $6M in annual ROI, improving customer operations by minin...