|By Michael Kopp||
|December 12, 2012 07:30 AM EST||
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We recognize that the average no longer represents the bulk of the transactions but can be a lot higher than the median.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.
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!
|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?
In 2006, Martin Fowler posted his now famous essay on Continuous Integration. Looking back, what seemed revolutionary, radical or just plain crazy is now common, pedestrian and "just what you do." I love it. Back then, building and releasing software was a real pain. Integration was something you did at the end, after code complete, and we didn't know how long it would take. Some people may recall how we, as an industry, spent a massive amount of time integrating code from one team with another...
May. 1, 2016 12:15 AM EDT Reads: 765
SYS-CON Events announced today that Peak 10, Inc., a national IT infrastructure and cloud services provider, 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. Peak 10 provides reliable, tailored data center and network services, cloud and managed services. Its solutions are designed to scale and adapt to customers’ changing business needs, enabling them to lower costs, improve performance and focus inter...
May. 1, 2016 12:00 AM EDT Reads: 1,050
As the software delivery industry continues to evolve and mature, the challenge of managing the growing list of the tools and processes becomes more daunting every day. Today, Application Lifecycle Management (ALM) platforms are proving most valuable by providing the governance, management and coordination for every stage of development, deployment and release. Recently, I spoke with Madison Moore at SD Times about the changing market and where ALM is headed.
Apr. 30, 2016 09:00 PM EDT Reads: 1,468
If there is anything we have learned by now, is that every business paves their own unique path for releasing software- every pipeline, implementation and practices are a bit different, and DevOps comes in all shapes and sizes. Software delivery practices are often comprised of set of several complementing (or even competing) methodologies – such as leveraging Agile, DevOps and even a mix of ITIL, to create the combination that’s most suitable for your organization and that maximize your busines...
Apr. 30, 2016 08:15 PM EDT Reads: 1,797
Struggling to keep up with increasing application demand? Learn how Platform as a Service (PaaS) can streamline application development processes and make resource management easy.
Apr. 30, 2016 08:00 PM EDT Reads: 2,053
The goal of any tech business worth its salt is to provide the best product or service to its clients in the most efficient and cost-effective way possible. This is just as true in the development of software products as it is in other product design services. Microservices, an app architecture style that leans mostly on independent, self-contained programs, are quickly becoming the new norm, so to speak. With this change comes a declining reliance on older SOAs like COBRA, a push toward more s...
Apr. 30, 2016 06:30 PM EDT Reads: 1,319
This is not a small hotel event. It is also not a big vendor party where politicians and entertainers are more important than real content. This is Cloud Expo, the world's longest-running conference and exhibition focused on Cloud Computing and all that it entails. If you want serious presentations and valuable insight about Cloud Computing for three straight days, then register now for Cloud Expo.
Apr. 30, 2016 02:30 PM EDT Reads: 1,710
SYS-CON Events announced today that Stratoscale, the software company developing the next generation data center operating system, 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. Stratoscale is revolutionizing the data center with a zero-to-cloud-in-minutes solution. With Stratoscale’s hardware-agnostic, Software Defined Data Center (SDDC) solution to store everything, run anything and scale everywhere...
Apr. 30, 2016 01:15 PM EDT Reads: 1,531
Digital means customer preferences and behavior are driving enterprise technology decisions to be sure, but let’s not forget our employees. After all, when we say customer, we mean customer writ large, including partners, supply chain participants, and yes, those salaried denizens whose daily labor forms the cornerstone of the enterprise. While your customers bask in the warm rays of your digital efforts, are your employees toiling away in the dark recesses of your enterprise, pecking data into...
Apr. 30, 2016 12:00 PM EDT Reads: 972
SYS-CON Events announced today that Men & Mice, the leading global provider of DNS, DHCP and IP address management overlay 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. The Men & Mice Suite overlay solution is already known for its powerful application in heterogeneous operating environments, enabling enterprises to scale without fuss. Building on a solid range of diverse platform support,...
Apr. 30, 2016 12:00 PM EDT Reads: 2,294
You deployed your app with the Bluemix PaaS and it's gaining some serious traction, so it's time to make some tweaks. Did you design your application in a way that it can scale in the cloud? Were you even thinking about the cloud when you built the app? If not, chances are your app is going to break. Check out this webcast to learn various techniques for designing applications that will scale successfully in Bluemix, for the confidence you need to take your apps to the next level and beyond.
Apr. 30, 2016 11:30 AM EDT Reads: 1,466
I had the opportunity to catch up with Chris Corriere - DevOps Engineer at AutoTrader - to talk about his experiences in the realm of Rugged DevOps. We discussed automation, culture and collaboration, and which thought leaders he is following. Chris Corriere: Hey, I'm Chris Corriere. I'm a DevOps Engineer AutoTrader. Derek Weeks: Today we're going to talk about Rugged DevOps. It's a subject that's gaining a lot of traction in the community but not a lot of people are really familiar with wh...
Apr. 30, 2016 09:45 AM EDT Reads: 1,570
Wow, if you ever wanted to learn about Rugged DevOps (some call it DevSecOps), sit down for a spell with Shannon Lietz, Ian Allison and Scott Kennedy from Intuit. We discussed a number of important topics including internal war games, culture hacking, gamification of Rugged DevOps and starting as a small team. There are 100 gold nuggets in this conversation for novices and experts alike.
Apr. 30, 2016 09:15 AM EDT Reads: 524
SYS-CON Events announced today that DatacenterDynamics has been named “Media Sponsor” of 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. DatacenterDynamics is a brand of DCD Group, a global B2B media and publishing company that develops products to help senior professionals in the world's most ICT dependent organizations make risk-based infrastructure and capacity decisions.
Apr. 30, 2016 06:00 AM EDT Reads: 2,467
With DevOps becoming more well-known and established practice in nearly every industry that delivers software, it is important to continually reassess its efficacy. This week’s top 10 includes a discussion on how the quick uptake of DevOps adoption in the enterprise has posed some serious challenges. Additionally, organizations who have taken the DevOps plunge must find ways to find, hire and keep their DevOps talent in order to keep the machine running smoothly.
Apr. 30, 2016 04:45 AM EDT Reads: 1,394
Between the mockups and specs produced by analysts, and resulting applications built by developers, there exists a gulf where projects fail, costs spiral, and applications disappoint. Methodologies like Agile attempt to address this with intensified communication, with partial success but many limitations. In his session at 18th Cloud Expo, Charles Kendrick, CTO & Chief Architect at Isomorphic Software, will present a revolutionary model enabled by new technologies. Learn how business and devel...
Apr. 30, 2016 04:00 AM EDT Reads: 1,730
Call it DevOps or not, if you are concerned about releasing more code faster and at a higher quality, the resulting software delivery chain and process will look and smell like DevOps. But for existing development teams, no matter what the velocity objective is, getting from here to there is not something that can be done without a plan. Moving your release cadence from months to weeks is not just about learning Agile practices and getting some automation tools. It involves people, tooling and ...
Apr. 30, 2016 04:00 AM EDT Reads: 1,491
The notion of customer journeys, of course, are central to the digital marketer’s playbook. Clearly, enterprises should focus their digital efforts on such journeys, as they represent customer interactions over time. But making customer journeys the centerpiece of the enterprise architecture, however, leaves more questions than answers. The challenge arises when EAs consider the context of the customer journey in the overall architecture as well as the architectural elements that make up each...
Apr. 30, 2016 02:45 AM EDT Reads: 1,948
Much of the discussion around cloud DevOps focuses on the speed with which companies need to get new code into production. This focus is important – because in an increasingly digital marketplace, new code enables new value propositions. New code is also often essential for maintaining competitive parity with market innovators. But new code doesn’t just have to deliver the functionality the business requires. It also has to behave well because the behavior of code in the cloud affects performan...
Apr. 30, 2016 02:30 AM EDT Reads: 1,352
APIs have taken the world by storm in recent years. The use of APIs has gone beyond just traditional "software" companies, to companies and organizations across industries using APIs to share information and power their applications. For some organizations, APIs are the biggest revenue drivers. For example, Salesforce generates nearly 50% of annual revenue through APIs. In other cases, APIs can increase a business's footprint and initiate collaboration. Netflix, for example, reported over 5 bi...
Apr. 29, 2016 09:30 PM EDT Reads: 2,561