Welcome!

AJAX & REA Authors: Alfredo Diaz, Andreas Grabner, Tim Hinds, RealWire News Distribution

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We recognize that the average no longer represents the bulk of the transactions but can be a lot higher than the median.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

Cloud Expo Breaking News
Web conferencing in a public cloud has the same risks as any other cloud service. If you have ever had concerns over the types of data being shared in your employees’ web conferences, such as IP, financials or customer data, then it’s time to look at web conferencing in a private cloud. In her session at 14th Cloud Expo, Courtney Behrens, Senior Marketing Manager at Brother International, will discuss how issues that had previously been out of your control, like performance, advanced administration and compliance, can now be put back behind your firewall.
Cloud scalability and performance should be at the heart of every successful Internet venture. The infrastructure needs to be resilient, flexible, and fast – it’s best not to get caught thinking about architecture until the middle of an emergency, when it's too late. In his interactive, no-holds-barred session at 14th Cloud Expo, Phil Jackson, Development Community Advocate for SoftLayer, will dive into how to design and build-out the right cloud infrastructure.
More and more enterprises today are doing business by opening up their data and applications through APIs. Though forward-thinking and strategic, exposing APIs also increases the surface area for potential attack by hackers. To benefit from APIs while staying secure, enterprises and security architects need to continue to develop a deep understanding about API security and how it differs from traditional web application security or mobile application security. In his session at 14th Cloud Expo, Sachin Agarwal, VP of Product Marketing and Strategy at SOA Software, will walk you through the various aspects of how an API could be potentially exploited. He will discuss the necessary best practices to secure your data and enterprise applications while continue continuing to support your business’s digital initiatives.
Cloud backup and recovery services are critical to safeguarding an organization’s data and ensuring business continuity when technical failures and outages occur. With so many choices, how do you find the right provider for your specific needs? In his session at 14th Cloud Expo, Daniel Jacobson, Technology Manager at BUMI, will outline the key factors including backup configurations, proactive monitoring, data restoration, disaster recovery drills, security, compliance and data center resources. Aside from the technical considerations, the secret sauce in identifying the best vendor is the level of focus, expertise and specialization of their engineering team and support group, and how they monitor your day-to-day backups, provide recommendations, and guide you through restores when necessary.
The revolution that happened in the server universe over the past 15 years has resulted in an eco-system that is more open, more democratically innovative and produced better results in technically challenging dimensions like scale. The underpinnings of the revolution were common hardware, standards based APIs (ex. POSIX) and a strict adherence to layering and isolation between applications, daemons and kernel drivers/modules which allowed multiple types of development happen in parallel without hindering others. Put simply, today's server model is built on a consistent x86 platform with few surprises in its core components. A kernel abstracts away the platform, so that applications and daemons are decoupled from the hardware. In contrast, networking equipment is still stuck in the mainframe era. Today, networking equipment is a single appliance, including hardware, OS, applications and user interface come as a monolithic entity from a single vendor. Switching between different vendor'...
You use an agile process; your goal is to make your organization more agile. What about your data infrastructure? The truth is, today’s databases are anything but agile – they are effectively static repositories that are cumbersome to work with, difficult to change, and cannot keep pace with application demands. Performance suffers as a result, and it takes far longer than it should to deliver on new features and capabilities needed to make your organization competitive. As your application and business needs change, data repositories and structures get outmoded rapidly, resulting in increased work for application developers and slow performance for end users. Further, as data sizes grow into the Big Data realm, this problem is exacerbated and becomes even more difficult to address. A seemingly simple schema change can take hours (or more) to perform, and as requirements evolve the disconnect between existing data structures and actual needs diverge.
SYS-CON Events announced today that SherWeb, a long-time leading provider of cloud services and Microsoft's 2013 World Hosting Partner of the Year, will exhibit at SYS-CON's 14th International Cloud Expo®, which will take place on June 10–12, 2014, at the Javits Center in New York City, New York. A worldwide hosted services leader ranking in the prestigious North American Deloitte Technology Fast 500TM, and Microsoft's 2013 World Hosting Partner of the Year, SherWeb provides competitive cloud solutions to businesses and partners around the world. Founded in 1998, SherWeb is a privately owned company headquartered in Quebec, Canada. Its service portfolio includes Microsoft Exchange, SharePoint, Lync, Dynamics CRM and more.
The world of cloud and application development is not just for the hardened developer these days. In their session at 14th Cloud Expo, Phil Jackson, Development Community Advocate for SoftLayer, and Harold Hannon, Sr. Software Architect at SoftLayer, will pull back the curtain of the architecture of a fun demo application purpose-built for the cloud. They will focus on demonstrating how they leveraged compute, storage, messaging, and other cloud elements hosted at SoftLayer to lower the effort and difficulty of putting together a useful application. This will be an active demonstration and review of simple command-line tools and resources, so don’t be afraid if you are not a seasoned developer.
SYS-CON Events announced today that BUMI, a premium managed service provider specializing in data backup and recovery, will exhibit at SYS-CON's 14th International Cloud Expo®, which will take place on June 10–12, 2014, at the Javits Center in New York City, New York. Manhattan-based BUMI (Backup My Info!) is a premium managed service provider specializing in data backup and recovery. Founded in 2002, the company’s Here, There and Everywhere data backup and recovery solutions are utilized by more than 500 businesses. BUMI clients include professional service organizations such as banking, financial, insurance, accounting, hedge funds and law firms. The company is known for its relentless passion for customer service and support, and has won numerous awards, including Customer Service Provider of the Year and 10 Best Companies to Work For.
Chief Security Officers (CSO), CIOs and IT Directors are all concerned with providing a secure environment from which their business can innovate and customers can safely consume without the fear of Distributed Denial of Service attacks. To be successful in today's hyper-connected world, the enterprise needs to leverage the capabilities of the web and be ready to innovate without fear of DDoS attacks, concerns about application security and other threats. Organizations face great risk from increasingly frequent and sophisticated attempts to render web properties unavailable, and steal intellectual property or personally identifiable information. Layered security best practices extend security beyond the data center, delivering DDoS protection and maintaining site performance in the face of fast-changing threats.
From data center to cloud to the network. In his session at 3rd SDDC Expo, Raul Martynek, CEO of Net Access, will identify the challenges facing both data center providers and enterprise IT as they relate to cross-platform automation. He will then provide insight into designing, building, securing and managing the technology as an integrated service offering. Topics covered include: High-density data center design Network (and SDN) integration and automation Cloud (and hosting) infrastructure considerations Monitoring and security Management approaches Self-service and automation
In his session at 14th Cloud Expo, David Holmes, Vice President at OutSystems, will demonstrate the immense power that lives at the intersection of mobile apps and cloud application platforms. Attendees will participate in a live demonstration – an enterprise mobile app will be built and changed before their eyes – on their own devices. David Holmes brings over 20 years of high-tech marketing leadership to OutSystems. Prior to joining OutSystems, he was VP of Global Marketing for Damballa, a leading provider of network security solutions. Previously, he was SVP of Global Marketing for Jacada where his branding and positioning expertise helped drive the company from start-up days to a $55 million initial public offering on Nasdaq.
Performance is the intersection of power, agility, control, and choice. If you value performance, and more specifically consistent performance, you need to look beyond simple virtualized compute. Many factors need to be considered to create a truly performant environment. In his General Session at 14th Cloud Expo, Marc Jones, Vice President of Product Innovation for SoftLayer, will explain how to take advantage of a multitude of compute options and platform features to make cloud the cornerstone of your online presence.
Are you interested in accelerating innovation, simplifying deployments, reducing complexity, and lowering development costs? The cloud is changing the face of application development and deployment, with enterprise-grade infrastructure and platform services making it possible for you to build and rapidly scale enterprise applications. In his session at 14th Cloud Expo, Gene Eun, Sr. Director, Oracle Cloud at Oracle, will discuss the latest solutions and strategies for application developers and enterprise IT organizations to leverage Infrastructure as a Service (IaaS) and Platform as a Service (PaaS) to build and deploy modern business applications in the cloud.
Hybrid cloud refers to the federation of a public and private cloud environment for the purpose of extending the elastic and flexibility of compute, storage and network capabilities, in an on-demand, pay-as-you go basis. The hybrid approach allows a business to take advantage of the scalability and cost-effectiveness that a public cloud computing environment offers without exposing mission-critical applications and data to third-party vulnerabilities. Hybrid cloud environments involve complex management challenges. First, organizations struggle to maintain control over the resources that lie outside of their managed IT scope. They also need greater infrastructure visibility to help reduce maintenance costs and ensure that their company data and resources are properly handled and secured.