Welcome!

SOA & WOA Authors: Elizabeth White, Trevor Parsons, Gigaom Research, Tad Anderson, Pat Romanski

Related Topics: SOA & WOA, Virtualization

SOA & WOA: Article

Monitoring Web Applications Using Hyperic

Monitoring enables us to anticipate the issues

Monitoring applications is an important aspect of IT industry. Huge investments are made in setting and maintaining the IT infrastructure be it cloud or physical. To ensure maximum performance across your business, problems have to be identified and resolved before they affect the users. Monitoring enables us to anticipate the issues and hence resolve them before they start appearing as problems to users. We need to have visibility into IT infrastructure to identify the issue origin. For example, by monitoring the applications proactively, we can figure out the root cause whether its poor performance is due to the problem at resource level or at application level or at any other place and take action before it starts impacting the user. It gives us more control and confidence on the application and hence helps us in meeting the service quality requirements.

Performance issues can arise due to problems at any level - machine/ application/ transaction. We should be able to monitor through all levels so that exact point of origin can be located. Most of the current solutions provide monitoring at any one particular level. For example, some cloud based infrastructure solutions provide monitoring at machine level only. Problem need not be necessarily to be at machine level. There could be problem at the application level like increased thread count, more number of active connections to valuable resources, etc or at transaction level like response time, request rate, etc. Thus, monitoring across different levels gives us visibility to know where the problem lies. Once the problem origin is known, we can take corrective actions to resolve it before it impacts the user experience.

There should be a solution to achieve the same to dig out the actual problem. We are going to deal with it in this article. We are going to see how to monitor different metrics (CPU utilization, thread count per minute, response time) at different levels. We will use Hyperic as our monitoring tool to measure the metrics. It is a systems monitoring, server monitoring and IT management software. It is based on server-agent model. We will have a look at the basic architecture of Hyperic as well.

Use-case scenario - Problem statement

Consider a case where a web application becomes famous and workload increases on it. This causes the application's performance to go down which results in slow loading of web pages. There could be many reasons for the degradation of performance like CPU utilization exceeds certain level, application logic is poor, less number of resources in infrastructure which are not able to sustain the load and many more. Issue could be at any level as discussed. Availability and performance marks end users' experience. This necessitates the monitoring of applications at different levels so that any unexpected behaviour could be avoided by taking corrective measures on time.

Deep dive into the solution
One of the most common complaints heard from end user is ‘website responds very slowly'. Poor performance of application can drift the user away from web application and leaves a bad impact on them. To find out the reason for poor performance of application, we have to measure the metrics at different levels to dig out the actual issue. We are programmatically going to monitor CPU utilization at machine level, used thread count per minute at application level and response time at transaction level.

Monitoring requires a reliable tool which can track the metrics periodically and update the application administrators.  There are lots of monitoring tools present in the market with varying features. Some of the most commonly used are Hyperic, Nagios, Cacti, Ganglia, etc.  Most of the tools in the market represent the metric values in visual form which helps the administrators to monitor and analyze any unexpected behaviour of application.

Hyperic is one of the leading monitoring tools in this arena. Hyperic comes in two flavors - open source (under GPL licensing) and enterprise edition (commercial license). It is based on server-agent model.  Following is a short description of Hyperic key components:

  1. HQ Agent: HQ agent is responsible for collecting data of the machine in which it is installed and send it to server. Each machine which is to be monitored should have an HQ agent running.
  2. HQ Server: HQ server consolidates the data sent by agents installed on different machines and persist it in the Hyperic database. HQ server maintains the inventory which keeps the information of all the resources and their monitored values.
  3. HQ Portal: HQ portal is a graphical user interface which gives complete information about the resources being monitored. User can add/ remove/perform control actions/set alarms from this portal. It is highly customizable with provision of choosing the metric to be collected and changing the collection intervals.
  4. HQ Web Service API: It has java based API called HQAPI. It helps in accessing the resources and their metrics programmatically.
  5. HQ Plugins: Hyperic has its own set of resource plugins for collecting metrics and other operations. It is also possible to build a new plugin to support additional functionality.

A large number of metrics is managed by Hyperic. In case of virtual environment, Hyperic agents are installed on virtual machines. We will use HQAPI to access the resources.

Consider a web application deployed on Apache Tomcat as application server. We will deploy famous - Jpetstore web portal on application server and then monitor the same.

Deploy web application on application server
Copy the jpetstore application under ${Apache_Tomcat_Installation_dir}/webapps and start tomcat. Try accessing - http://localhost:8080/jpetstore link and if it shows jpetstore homepage that means application is deployed successfully.

The way resources are modelled in hierarchical fashion in Hyperic UI console, same way resources could be accessed programmatically using HQ Java API. Top most resource is Platform and beneath it are present services running which are allowed to be monitored in that machine. Let's see how to monitor different metrics at different levels using HQAPI.

More Stories By Akansha Jain

Akansha works as a Technology Analyst at SETLabs, R&D division, at Infosys Technologies Ltd. She has close to 4 years of experience in development of Cloud computing, Java and Java EE applications, Eclipse Plugin Architecture, Software Factory, Web 2.0,etc. Earlier publication works are available on devx.com.