Inexpensive Web Hosting - J2EE Hosting Sun Java, Mysql5, Java, Jsp, Servlet Developing Blog

CHAPTER 10 JAVA EE PERFORMANCE ASSESSMENT Examine access logs or your end-user experience monitor and analyze usage patterns against load scripts. Run a performance assessment to discover the actual end-user experience and the behavior of environment resources. It is only by combining the information gathered from user logs with the actual behavior of your environment that you can truly feel comfortable with the accuracy of your capacity assessment. In summary, the benefits performance assessments offer are tuning the performance of your application and validating that your tests and expectations are in line with reality. Performance Assessment Overview This section outlines the performance assessment stages: prerequisites, process, and analysis. Prerequisites Before starting a performance assessment, you need to satisfy some prerequisites for your assessment to be valuable: Formally defined SLA: You need a formally defined SLA to accurately assess if a request is performing as expected. Without a formally defined SLA, your assessment is essentially a stab in the dark the best you can do is assume that because a service request is taking a long time to run that it is problematic, without knowing how the request is supposed to respond. Application monitoring: You need to implement application monitoring in the form of bytecode instrumentation to isolate performance problems. It is not enough to learn that your application or even a specific request is not performing as expected, unless you can properly isolate the cause of the performance problem. Application server monitoring: Application server monitoring identifies the performance of application server resources such as heap management and garbage collection, thread pools, and connection pools to enable you to identify if application server tuning can help the performance of your application. Resource monitoring: Outside the application, you need to monitor the performance of operating system resources such as server CPUs, disk I/O rates, and network communications, in addition to external resources such as databases and legacy systems. Without this level of monitoring, application performance anomalies may be miscategorized as application code issues. Preproduction requirements: If you are implementing a performance assessment against a preproduction or production staging environment, then you need a representative subset of production as well as appropriate load to test your application against. You can only tune an application to the way that your users use it; if your load is not balanced and representative of actual end-user behavior, then you cannot have confidence that you have identified performance bottlenecks.
We recommend you use shared web hosting services, because many users agree that it is cheap, reliable and customer-satisfying webhost.

256 CHAPTER 10 JAVA EE PERFORMANCE ASSESSMENT While a performance assessment is typically conducted against a production environment, if balanced and representative load can be generated against a production staging environment, then it can be conducted there. The difference between recording information from a production staging environment and recording from a production environment is that when recording information from a production environment, a strategy to minimize the impact on production environment end-users must be employed. Consider that you always affect a system by observing it, and the impact of observing the system needs to be mitigated so that your users do not experience a performance degradation and you do not affect the system so much that it skews your observations (for example, by producing false positives). In the case of the aforementioned telecommunications company, we implemented a staged approach to capture live production data with minimal impact on the end-user experience. In this chapter, we discuss the reasons for and strategies behind building a Java EE performance analysis report. We start off by taking a closer look at the benefits of performance assessments. Performance Assessment Benefits Why do we need to run Java EE performance assessments? Isn t a capacity assessment enough? What benefits do performance assessments offer? These common questions I hear from customers when I help them implement a full performance management plan are quite valid: why should they dedicate time and resources to an activity unless they can realize substantial profit from it? To address these questions, you need to assess the health of your enterprise Java environment for the following reasons: You cannot make any assumptions about the capabilities of your enterprise Java environment until you have tested it. As the saying goes, an ounce of prevention is worth a pound of cure by assessing your environment prior to deployment, you can prevent unwanted issues from arising after the application goes live. You cannot have confidence in your capacity assessment until you have validated it. The first point is straightforward: have you ever developed and deployed an application without testing its performance? You can develop an application with appropriate design patterns and best practices, but until you subject it to a load test while monitoring it with byte- code instrumentation, all you can do is hope for the best. Rather, you need to identify and resolve application, platform, and dependency bottlenecks before deploying the application. I apologize for the clich in the second point, but it is 100 percent true with respect to application performance: spending one week performance testing your application and identifying and resolving bottlenecks prior to deployment is better than spending one hour resolving a production performance issue while your users wait. A performance assessment is the formal mechanism employed to guide tuning efforts. Regarding the third point, when your application moves from a production staging environment into production, your capacity assessment is put to the test. Considering that you are required to build your capacity assessment from a load testing tool and a set of load scripts, you still require validation that your load tests accurately represent actual end-user behavior. Two simultaneous approaches to validating the accuracy of your capacity assessment are as follows:
Note: If you are looking for cheap and reliable webhost to host and run your mysql application check mysql web server services.

CHAPTER 10 Java EE Performance Assessment Now that our application is in production and not crashing, I can finally relax, John sighed. It had been several months of hard work rebuilding his application and environment. Not quite yet, I replied. Now you need to validate that your environment is in fact behaving as you expect it to. So what do you recommend that I do? John queried. I recommend that you periodically perform a Java EE performance assessment. This assessment will reveal the performance of your application requests, your environment configuration, and your external dependencies. But how is that different from a capacity assessment? Well, a capacity assessment is done to, among other things, determine how much load will cause your environment to break, I responded. A performance assessment is done to determine how your application is performing in your production environment. You wouldn t want to break production now, would you? This is the key difference between the purposes of a capacity assessment and a performance assessment: a capacity assessment identifies the limitations of an environment, whereas a performance assessment identifies the current performance of an environment. Of course not. What do I need to get started? asked John. You are going to need some monitoring tools and a strategy to gather performance information from those tools without disrupting your end users experience. Let s get started. Whereas a capacity assessment is performed rather infrequently, such as at the conclusion of a significant iteration or just prior to the production rollout of a new version of an application, a Java EE performance assessment is performed frequently as part of a proactive plan to improve the performance of your applications. The purpose of a performance assessment is to identify performance bottlenecks or simply as a mechanism to proactively tune an application. For example, I once worked with a customer in the telecommunications industry to set up a weekly process to identify the company s top five slowest SQL statements and top five slowest service requests. The person I worked with managed the system administration group, which in this particular organization included database administrators, and he wanted a constant set of tasks that his team could be working on when not troubleshooting issues. Whenever members of this group had breaks in their work schedules, they would start working on bringing down their top fives.
We highly recommend you visit web and email hosting services if you need stable and cheap web hosting platform for your web applications.

PART 3 Performance Management in Production PART 3 Performance Management in Production
Note: If you are looking for cheap and reliable webhost to host and run your mysql application check mysql web server services.

CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING The primary factor impeding the performance of the application is the CPU utilization of servers running in the application server tier. Therefore, either the addition of CPUs to existing servers in this tier or of a new physical machine is recommended. Furthermore, the latest version of the application uses and saturates the CPU faster than the previous version; therefore, analyzing in a code-profiling tool the use cases identified as being problematic in the use case degradation model is recommended to determine the root of the performance changes. Summary This chapter discussed the difference between the concepts of performance and scalability: performance measures the speed with which a single request can be executed, while scalability measures the ability of a request to maintain its performance under increasing load. It also outlined the strategy of ensuring performance before testing for scalability. This strategy led to a detailed exploration into the ultimate scalability test the capacity assessment. A capacity assessment identifies the following key points about your environment: Its performance at expected load The load that causes it to violate its SLAs The load that causes the environment to become saturated The load that forces the environment into the buckle zone Equipped with the correct load testing strategy and monitoring tools, we explored how to ascertain this information and assemble it into a formal Capacity Assessment Report. In the next chapter, we explore performance assessments that occur more frequently and are used to diagnose performance issues and validate the accuracy of capacity assessments.
We highly recommend you visit web and email hosting services if you need stable and cheap web hosting platform for your web applications.

250 CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING Figure 9-11. The aggregate CPU utilizations at the application server tier between versions 1.1 and 2.0 of the Buy High, Sell Low application demonstrate that version 2.0 makes heavier use of its CPUs at lower user load than version 1.1. Note Many times the code functionality between major versions of an application is significantly different, so a deterministic impact analysis is impossible to perform. In these cases, performing the impact analysis is still beneficial, but focus on comparing the response times and resource utilizations of the expected usage of the new version against the response time and resource utilizations of the expected usage of the previous version. Figure 9-10 plots the average response time buffers rather than request response times to focus on that comparison the average response time buffer value is relative to each use case s SLA. You may need to add this disclaimer to your impact analysis: The impact analysis reports the differences between the performance of an enterprise environment executing two sets of application code functionality against their individual performance criteria; it does not necessarily reflect a direct response time impact. Final Analysis and Recommendations At the current user load of 500 users, all use cases are satisfied, and the environment can support an additional 50 users. This represents a 10 percent user buffer, which by industry standards is too low: the optimal user buffer for a volatile application like the Buy High, Sell Low application is anywhere between 25 and 40 percent.
Please visit Domain Name Hosting services for high quality webhost to host and run your jsp applications.

CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING Figure 9-10 illustrates the impact of the version 2.0 code against the version 1.1 code. Figure 9-10. The response time impact analysis The growth patterns between the two response time buffers are similar, but version 2.0 code crosses the SLA violation point at 550 users while the version 1.1 code crosses the SLA violation point at 650 users. This pattern shift represents a degradation in overall application capacity of 15.4 percent. The resource utilization of both code versions was CPU-bound, but version 2.0 code degraded sooner than the previous version, as illustrated in Figure 9-11. Continue to add resource degradation graphs for relevant resources in your capacity analysis, and where relevant, include graphs illustrating the performance differences between unchanged use cases. For example, if the login functionality did not change between versions, and the performance degraded, then calling out resource utilization differences in the impact analysis is important. However, if the functionality dramatically changed, such as swapping a text file for an LDAP server for user validation upon login, then a direct comparison of response times is not particularly useful to include (unless, of course, this comparison was requested). When functionality changes significantly, the best option is to compare the response time buffers between the application versions: is the new login functionality (with its new SLA) satisfying its SLA as well or better than the previous version?
In case you need quality webspace to host and run your web applications, try our personal web hosting services.

248 CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING Application Tier Insert graphs of the application server machines and instances, including the following information: CPU utilization Physical memory utilization Process memory utilization Application server instance heap utilization and garbage collection rates Application server instance thread pool utilizations Application server instance pool utilizations Application server instance cache counts of activation/passivation, hits, and misses Database Tier Insert graphs of the database machine and performance information, including the following information: CPU utilization Physical memory utilization Process memory utilization Disk I/O rates Cache hit/miss counts Network Insert graphs that display the response time and load between all servers against user load. Impact Analysis The performance of the Buy High, Sell Low application has degraded with the release of version 2.0. The average response time degradation is 12 percent, and the average resource utilization at expected load has increased by 7 percent. Throughput at expected usage has likewise degraded by 10 percent. The maximum capacity has decreased from 650 users to 550 users, a degradation of 15.4 percent.
We recommend high quality webhost to host and run your jsp application: christian web host services.

CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING Figure 9-9. The response time for the Login use case Resource Degradation Model The Buy High, Sell Low application is CPU-bound primarily on the application server tier. Figure 9-8 shows the aggregate CPU utilization for the two machines running in the application server tier. This section details the performance of each component in each tier as observed during the capacity assessment. Note This section, the largest in the Capacity Assessment Report, contains performance graphs for all physical machines, internal servers, and network interactions. Pick the resources that are relevant for your environment and combine multiple related metrics on the same graph. The point of this section is to illustrate the degradation of resources over user load, and to record the behavior of resources for historical impact analysis. Therefore this section reports resources that degrade as well as resources that do not. Web Tier Insert graphs of the Web servers, including the following: CPU utilization Physical memory utilization Process memory utilization Request throughput Thread pool utilization
We recommend high quality webhost to host and run your jsp application: christian web host services.

246 CHAPTER 9 PERFORMANCE AND SCALABILITY TESTING Figure 9-8. The application server tier CPU aggregate Use Case Degradation Model This section presents the performance of each use case performed during the capacity assessment. Use Case: Login Figure 9-9 illustrates the performance of the Login use case throughout the capacity assessment. The response time for the Login request stayed relatively steady below the four-second mark until reaching about 500 users; it experienced a couple spikes above four seconds up until this point, but nothing sustained or in violation of the flexibility of the use case s SLA. Consistent with the observed degradation patterns, after the user load exceeded 550 users, the response time started to increase significantly.
From our experience, we are can tell you that you can find a reliable and cheap webhost service at Java Web Hosting services.