Active vs. Passive Web Performance Monitoring

Executive Summary

This paper examines the website monitoring options available with passive and active monitoring solutions. Passive monitoring is better suited as a technology to determine what went wrong with a website or Web application after problems have occurred. Active monitoring solutions allow the webmaster to identify problems, such as application issues and network or server connectivity problems, and to determine if a website or application is experiencing downtime or performance degradation before these problems impact a large number of site visitors.

Introduction

The Internet has evolved into a ubiquitous global technology that truly shrinks geographical barriers. Whether it’s a two-person company operating from a kitchen table to a multinational conglomerate with offices around the world, the Internet levels the playing field. A well-designed website is a great equalizer. But just as important as building and deploying a first-class website is managing the website. Website monitoring is critical to the online success of a business. Because the Internet is not without risk and the complexity of current Web applications is ever-increasing, it is critical to safeguard the integrity of a website in terms of end-user accessibility, availability, and performance. There are simple solutions available that send a single ping request to check network connectivity, and there are more comprehensive solutions that conduct a detailed site analysis at regular intervals. The key is to understand what is needed and what factors to evaluate.

Neglecting a Website Costs Money

Many companies take a hands-off approach to their website. This is a critical mistake. Whether it’s an e-commerce or a purely informational website, the implications of site outages are significant. Customers demand that a website consistently deliver a steady performance. Without this reliable, consistent performance of a website, customers will seek better performing and optimized websites to patronize.

E-commerce sites get quite a bit of attention in the media. One of the largest and most well-known e-commerce sites is Amazon.com. They have survived the dot-com bust, extreme cut-throat competition, infrastructure demands, and more recently, the weak economy. Amazon bets their multibillion-dollar business on their website running smoothly and staying up virtually all of the time. In June 2008, Amazon.com went down for almost two hours. The cost of the outage was unclear at the time because it only affected certain geographies. However, the company reported income of more than $4 billion¹ during the quarter that included June 2008, so on average, it booked $1.8 million in sales every hour. Numbers like these make even the briefest of outages significant.

While this is a high-profile example of a large online retailer, the effects on other companies are just as pronounced. Whatever the business—from consulting firm to real-estate agency to air-conditioning repair—website problems correlate directly to lost revenues and reputation.

Monitoring Is Essential

While there is a myriad of solutions for Web monitoring, there are two primary types of solutions for monitoring Web infrastructure:

• Passive monitoring
• Active monitoring

Passive monitoring takes an internal approach to monitoring the performance of Web applications from within the enterprise network. Active monitoring, also referred to as synthetic monitoring, performs regular, scripted checks that simulate end-user behavior. These tests can be run from inside the network or from multiple points on the Internet. Running these checks externally replicates an end-user experience as closely as possible. While passive monitoring and active monitoring aim to provide critical performance monitoring data, the methodology, and thus the result, is completely different.

Passive Monitoring

Passive monitoring, also known as real-user monitoring, is commonplace in large enterprises, where IT departments use enterprise management tools to monitor application performance within its network. Typically, these tools are hardware based and sit inside the LAN capturing traffic flow as it enters and leaves the site. Passive website monitoring collects relevant metrics in response to outside stimuli, with the goal of monitoring real users on a website and their behavior on the site. Passive monitoring technology is ideally suited to monitor actual end-user interactions as they happen. Passive monitors are available at a wide range of price points and implementation methods. These are hosted solutions, found in third-party data centers, while other solutions are hosted locally, most often on the Web server itself or as part of a hardware appliance.

Some monitors focus on client-side experience through the use of a beacon, while others target server-side performance through instrumentation on the Web server or using a network port. Passive monitoring has several advantages and disadvantages.

The advantages to client-side passive monitors are that they allow the following:

• Monitoring of requests that may be cached by the user agent or a proxy
• Monitoring of client-side events and activities
• Application spanning
• Monitoring of page-to-page latency as seen by the user
• Monitoring actual page render time

There are two main disadvantages in passive monitoring:

• It requires real end-user traffic; if users are not interacting with the site, the passive monitors will have no data to process.
• It offers no ability to monitor issues that occur outside of the network, such as DNS problems, network issues, and firewall and router outages.

While passive monitoring has some nice capabilities, it is better suited as a troubleshooting technology for determining what went wrong with a site after the problem has occurred.

Active Monitoring

Active monitoring uses technology that emulates a Web browser. Scripting is programmed to simulate the actions of an actual user of the website. These scripts execute continuously. As they execute, they record and analyze multiple metrics. The reportable data includes website availability and performance.

Active monitoring can detect performance and connectivity issues from multiple Internet points. This is an external approach and thus is designed to replicate an end-user experience. Active monitoring solutions enable problem identification—it can determine if a website or application is slow or experiencing downtime within minutes of the occurrence. Furthermore, by monitoring from an external perspective, active monitoring enables pinpointing the cause of the issue, such as slow applications, Internet network problems, and infrastructure issues. Since active monitoring solutions do not rely on real Web traffic, applications can be tested continually, providing up-to-the-minute status on website availability and performance.

There are a number of benefits realized with active performance monitoring:

• It is proactive, so performance issues can be detected before they affect websites and Web applications.
• Monitoring can occur from multiple Internet locations, allowing for a more holistic view of end-to-end connectivity for geographically distributed users.
• The ability to simulate user transactions allows for performance monitoring of scripted business processes, such as the use of shopping carts, secure log-ins, forms, etc.
• It allows for more relevant service-level agreement management by providing metrics from an end-user perspective.

Best-of-breed active monitoring solutions should include the following:

• Extensive reporting for uptime and performance using a custom definable matrix
• A flexible alerting infrastructure to deliver notifications reliably to multiple points
• Escalation mechanisms to manage notifications efficiently within multiple support units
• Content monitoring to ensure there are no errors within the page, including checking referenced objects, SSL certificates, and page content
• Easy-to-use transactional monitoring, with macros to capture step-by-step business processes
• Geographically distributed monitoring locations to access websites and Web applications using different Internet backbones
• Page snapshots and network trace routes when problems do occur, allowing for root-cause analysis

Active monitoring is easy to set up and easy to use. Services such as Dotcom-Monitor^® (www.dotcom-monitor.com) enable a website owner to set-up monitoring for both HTTP and HTTPS and check availability as often as once-per-minute from monitoring stations around the globe. Most active monitoring providers deliver their services using the software-as-a-service model, so there is no hardware or software investment required. Non-technical users who know how to use a Web browser can easily set up monitoring scenarios and then receive alerts and reports whenever a performance issue occurs.

What’s the Right Call?

Both passive and active monitoring offer benefits, and some of these are not mutually exclusive. While both active and passive solutions aim to provide critical performance monitoring data, the methodology and the results are completely different.

In reality, an approach that combines both passive and active monitoring solutions offers the highest degree of performance assurance. Performance issues must be detected before they occur, but problems occur even in the best scenarios. A combination of active, external monitoring and passive, behind-the-firewall monitoring offers the best solution to decrease unplanned downtime.

Lastly, for companies that choose to outsource their hosting, third-party active monitoring offers a way to validate SLA information provided by the hosting company. Solely depending on reports from a host provider is a head-in-the-sand approach to managing the relationship with the vendor and the availability and performance of the corporate Web presence. Third-party active monitoring allows for proactive management of a strategic asset.

¹ Source: Form 10-Q from Amazon Com Inc., filed on July 25, 2008 with the U.S. Securities and Exchange Commission.