eG Total Performance Visibility


Web Server Monitoring


Web Server Performance Monitoring Growing in Importance

Web servers like Microsoft IIS, Apache, IBM HTTP server, Oracle HTTP server and Sun ONE web servers are the heart of IT infrastructures in various domains - Healthcare, Banking, Trading, Logistics, etc. To ensure scalability and high performance, most web sites are being architected to use the multi-tier model - i.e., with the web server (IIS, Apache, etc.) functioning as the front-end, the middleware application server (J2EE, .Net based, etc.) that hosts the business logic functioning as the mid-tier, and a database server (SQL, Oracle, etc.) as the backend. In such architectures, the web server plays a pivotal role since all users to the other tiers are routed via the web server and hence, any slowdown or problem in the web server tier can adversely impact the end user experience.

The availability of a web site and the response time for user accesses to the site are the most critical metrics of web server performance. Both these metrics may vary depending from one website to another and even from one transaction to another. For instance, one set of application components may come into play when a user logs in to an eBanking site, while a set of components may be invoked when a user transfers funds between his/her accounts. Consequently, a web server performance monitoring solution must be able to report the availability and response time for individual user transactions to a web site.

divider

Active Monitoring of Web Server Performance

Most web server monitoring software solutions rely on request emulation to monitor web transactions to a site. These request emulators generate synthetic requests periodically from one or more locations to the site and monitor the availability and response time for each transaction. This simple yet elegant solution provides the external perspective of the site. The main limitations of a request emulation-only approach are:
  • This approach cannot be used to monitor the most critical transactions to a web site e.g., a user making a payment, a user registering to a web site, etc.
  • Moreover, this approach mainly samples the functioning of the target environment. If a specific transaction is failing, say 10% of the time, the emulation approach only has a 10% chance of reporting the problem. Consequently, this approach is able to consistently detect and report problems only when they are severe enough to impact the end user performance. That is, a request emulation approach only enables reactive monitoring.
divider

Combined Active and Passive Monitoring of Web Server Performance

The eG web server monitoring software adopts a unique two-pronged approach for web transaction monitoring. The external agent uses request emulation to assess the user experience from different locations. By doing so, the external agent captures the effect of the network latency and the server-side processing time on the end user experience.


Web Server Monitoring Software
An IIS monitoring example showing monitoring of web transactions

To quantify the server processing times for real user requests (not emulated requests), the eG internal agent deploys a proprietary web-adapter technology. This technology enhances web servers with the capability to track HTTP/HTTPS requests to a web server and the corresponding responses. For each transaction that is configured for monitoring, the web adapter analyzes the request URLs and responses to report various metrics relating to individual web transactions in real-time.


Server Performance Monitoring
Web monitoring and diagnosis: Comparison of response times across tiers to identify performance bottlenecks

The server performance monitoring is done in an implementation-independent manner, as a result of which eG agents are able to monitor Java (Servlets, EJB, JSPs) and other non-Java implementations (ASP, PHP, CGI, etc.) with equal felicity. Since it is able to monitor real-user transactions to web servers in real-time, eG Enterprise's web adapter technology enables the agents to proactively monitor and quantify all anomalies that may occur in a web infrastructure. The ability to offer real-time monitoring of 100% of the real user transactions, without the need for explicit, expensive logging is a key distinguishing feature of eG Enterprise's Web Server Monitor. This one-of-a-kind monitoring capacity supports Microsoft IIS, Apache, Sun ONE and other popular web servers.

divider

What the eG Web Server Monitor Reveals

Monitoring Capabilities What the eG Web Server Monitoring Reveals
External monitoring Is the web site available for user accesses from different locations? What is the response time for user accesses to the site from different geographic locations? Is a slowdown due to increased network latency or due to increased server-side processing?
Internal transaction monitoring How are the critical transactions of a web site functioning? What is the request rate for each transaction? What is the average response time for each transaction? Are there many aborts for the transaction?
Web site monitoring What is the status of the different web sites hosted on a single web server? Are there many errors occurring in the system? Are the servers supporting the web infrastructure adequately sized? Are there usage trends that need to be accounted for future capacity planning?
Bottleneck detection Is an increase in server-side processing time due to the web server or due to the middleware application server or due to the database?
Capacity planning Is the load being effectively balanced across all the web servers? Are the critical web server processes up and running?

Benefits of the eG Web Server Performance Monitoring Tool

  • Obtain deep insights into the TRUE end user experience
  • Receive proactive, real-time alerts of server availability and performance problems and get to know of any anomalies before they impact the business service performance
  • Analyze and correlate web site performance in context - taking into account the network, middleware and database performance
  • Historical analysis and trending of web services usage enables effective capacity planning