Performance Analysis FIG explained Loadrunner
Transactions (user transaction analysis)
user transaction analysis is the basis for performance analysis of a user standing on an angle.
1, Transation Sunmmary (transaction review)
transaction is the first step in a comprehensive analysis of performance analysis, by analyzing the successes and failures within the test time user transaction can directly determine whether the system is functioning properly.
2, Average Transaciton Response Time (transaction average response time)
"Transaction average response time" shows the average time each second during the test scenario run transaction execution used, through which can analyze the performance to application system during a test scenario run .
Example: With the change of test time, transaction processing speed of the system began to slow down, which illustrate the application of the system changes with time of commissioning, the overall performance will be a downward trend.
3, Transactions per Second (by number of transactions per second / the TPS)
"by the number of transactions per second / TPS" seconds of the scene displayed in each operation, each transaction, and the number of failures is stopped, so that a system performance test Important parameters. It can be determined by the system time transaction load at any given moment. TPS analysis mainly to see the performance curve trend.
It will be the average transaction response time by comparing the number of transactions you can analyze the impact on the execution time.
Example: When the pressure increased, CTR / TPS curve changes slowly or if there is a flat trend is likely to start the server bottleneck.
4, Total Transactions per Second (by the total number of transactions per second)
"by the total number of transactions per second" in the scene is running, the total number of transactions each second passed, failed to stop Affairs Department and the total number of transactions.
5, Transaction Performance Sunmmary (Transaction Performance Summary)
"Transaction Performance Summary" display scheme smallest of all transactions, maximum and average execution time, response time can directly determine whether the user's requirements.
Focus affairs average and maximum execution time, if it is within the acceptable range not in the user's time, the need for cause analysis.
6, Transaction Response Time Under Load (transaction response time and load)
"Transaction Response Time and load" combination "running virtual user" FIGS "average response transaction time" chart, through which can be seen at any point in time transaction response time relationship with the number of users, so that the user know the performance of data concurrency in the system to provide a reference for the expansion of the user's system. This figure can view the virtual user load overall impact on the execution time, more useful for analysis of test scenarios with a gradual load.
7, Transaction Response Time (Percentile) ( transaction response time (percentage))
a comprehensive analysis map "Transaction Response Time (percentage)" is analyzed based on the test results obtained, which is the tool by chart some statistical analysis of indirect get. It may be analyzed by a percentage of the transaction response time can be performed within a given transaction.
8, Transaction Response Time (Distribution) ( transaction response time (distribution))
"Transaction Response Time (distribution)" is displayed in the scene during the operation, the transaction execution time spent distribution, which can understand the transaction testing different response time number. If the pre-defined transaction related minimum and maximum acceptable response time for the transaction, you can use this graph to determine whether the server performance in the acceptable range.
Web Resources (Web Resource Analysis)
Web resource analysis is to start from the server to the Web server performance analysis.
1, Hits per Second (clicks per second)
"Hits per Second", even running the scenario during the second virtual user submitted to the Web server's HTTP requests.
It can be assessed by the amount of virtual load user-generated, and as its "Average Transaction Response Time" comparison chart, you can see the impact of the number of clicks generated on transaction performance. By the stability of view "clicks per second," you can determine the system. The system typically indicates a lower CTR in the slow response speed of the server, need further analysis, we found that the system bottleneck.
2, Throughput (throughput)
"throughput" is displayed per second throughput scenario during the operation of the server. Which measure in bytes, represented by any virtual data amount per second is obtained from a given server.
Can be assessed based on the throughput of server load virtual user-generated, and the processing power of the server to see in terms of traffic and the presence of bottlenecks.
Difference map "throughput" chart and "hits":
"throughput" chart, is the number of HTTP server per application processed.
"CTR" chart, is the total amount of data per second to get the client from the server.
3, HTTP Status Code Summary (HTTP status code Outline)
"Summary HTTP status code" shows the number of HTTP status codes scenario or session-step process is returned from the Web server, the code of the packet according to FIG. HTTP status code represents the status of HTTP requests.
4, HTTP Responses per Second (HTTP Responses per Second)
"HTTP Responses per Second" is a process running scene from a number of different HTTP status code returned by a Web server, but also other types of second return status information codes by analyzing the status code, the server can judge the operation under pressure, can also be grouped by the results shown in the drawings, so as to locate the error code is generated script.
5, Pages Downloader per Second (number of pages downloaded per second)
"The number of pages downloaded per second" shows the number of pages per second is downloaded from the server scenario or session step run. Using this map may be calculated based on the number of pages downloaded Vuser load generated.
And FIG throughput as the number of icons is downloaded pages per second amount of data received from the server Vuser in any given second. However, the throughput of each resource considered extremely size (for example, the size of each GIF file, the size of each page). The number of pages per second download only consider the number of pages.
NOTE: To view the downloaded pages per second map, you must set the "number of pages per second (HTML mode only)" RTS there.
6, Retries per Second (retry number of times per second)
"Retry times per second" shows the number of connection attempts to each second scenario or session server running step.
In the following situations server connection is retried:
A, initial connection unauthorized
B, proxy authentication requires
C, the server closes initial connection
D, the initial connection to the server can not connect
E, the first server can not resolve the IP address of the load generator
7, retries Summary (retries Summary)
"retries Summary" display frequency and the step of connecting the scene during the operation or session server attempts, according to which the packet retry reasons. Process may be determined using a scene or session with a server running this step in FIG retry number of times per second retry diagrams which point in time.
8, Connections (connections)
"Connections" shows the number of TCP / IP connection or session scene during the step operation each time point opened.
With this map, you can know when you need to add other connections.
Example: When the number of connections increases rapidly reach a steady state and the transaction response time, add a connection can greatly improve the performance (to reduce transaction response time).
9, Connections Per Second (connections per second)
"connections per second" shows the number of TCP / IP connections established during operation programs per second.
Ideally, a lot of HTTP requests should use the same connection, rather than each request to open a new connection. FIG via connections per second can be seen that the processing of the server, it indicates that the server's performance gradually decreases.
10, SSLs Per Second (SSL connections per second)
"SSL connections per second" and displaying the new scene number SSL connection or session running steps each second open reused. When you open TCP / IP connection to a secure server, the browser will open an SSL connection.
Web Page Breakdown (page elements subdivision)
"page elements subdivision" is mainly used to assess whether the page content affect the response time of the transaction, through which you can download in-depth analysis of those graphic on the site is slow or interrupted connections in question
element.
1, Web Page Breakdown (page decomposition General Plan)
"page decomposition" shows the response of a particular transaction in the testing process, and then analyze matters related to running correctly.
"Page decomposition" may be further subdivided according to FIG following four ways:
. 1), the Download Time Breaddown (download time division)
"Download Time Segment" FIG display page download time of the different elements, but can also follow the download process the decomposition time, with different colors to display DNS resolution time, time to establish a connection, the first time the proportion of each buffer.
2), Component Breakdown (Over Time ) ( Component Breakdown (change with time))
"Component Breakdown" diagram shows the selected subdivision map web page components over time. From this figure we can easily see which elements unstable download time during the test. This figure is particularly suitable for the client needs to download controls more pages, response time analysis controls, it is easy to find those controls that are unstable or time-consuming.
3), Download Time Breakdown (Over Time) ( Download Time Breakdown (time-varying))
"Download Time Breakdown (Over Time)" graph shows the selected web page elements download time segments (changes over time) situation it shows very clearly the downloads page of each element in the stress test process.
"Download Time Breakdown," chart shows the time throughout the test page elements in response to the results of statistical analysis, "Download Time Breakdown (Over Time)" process what the scenario is displayed in the page elements within the response time statistics every second As a result, both are from the perspective of macro and micro analysis download time of page elements.
4), Time to First Buffer Breakdown (Over Time) ( First Buffer Breakdown (time-varying))
"First Buffer Breakdown (time-varying)" chart shows success returns received from the Web server the period before the first buffer, the server each time the page every second component scene or session and the step of running network time (in seconds). This may be used to determine FIG scene or session of time steps during the operation of the server or network.
First Buffer Time: refers to the client to establish a connection with the server, the first packet transmission start time from the server via the data network to the client, to the browser receives the first time a buffer used.
2, Page Component Breakdown (Page Component Breakdown)
"Page Component Breakdown," chart shows the average download time for each page and its components (in seconds). FIG column can be sorted according to the average number of seconds download a component through which help isolate the component in question.
3, Page Component Breakdown (Over Time ) ( page exploded assembly (time-varying))
"page exploded assembly (versus time)" graph shows the average response time per page and its components during the operation of each second embodiment (in seconds).
4, Page Download Time Breakdown (Page Download Time Breakdown)
"Page Download Time Breakdown" diagram shows the breakdown of each page component's download time can be determined according to its website during the download transaction response time is slow or caused by a network error errors caused by the server.
"Page download time segment" according to FIG DNS resolution time, connection time, time to first buffer, the SSL handshake time, reception time, FTP verification time, and the client time error time for the download process to segment each component .
5, Page Download Time Breakdown (Over Time) ( page download time segments (time-varying))
"page download time segments (time-varying)" FIG display during the scenario, each second download time of each page component the subdivision. FIG determine this problem occurs in the network or server at what point during the execution of the program.
"Breakdown (Over Time) page components" chart and "Page Download Time Breakdown (Over Time)" chart is usually combined for analysis: first determine the component in question, and then analyze their download process, so as to locate the reason where.
6, Time to First Buffer Breakdown (First Buffer Breakdown)
"First Buffer Breakdown," chart for each page component within this period of time before the first buffer is successfully received back from the Web server related server / network time. If the download time component is very long, you can use this graph to determine issues arising from or related to the server and network related.
Network Time: defined as the first HTTP request that moment, the average time that elapsed until confirmed.
Server Time: defined as from the receipt of the initial HTTP request confirmation begin until successfully received the average time from the Web server that elapsed between the first buffer.
7, Time to First Buffer Breakdown ( Over Time) ( First Buffer Breakdown (time-varying))
successfully received the first returned from a Web server "First Buffer Breakdown (Over Time)" graph shows During this time before a buffer, each time the server component of the second page of each operation scene and network time. You can use this chart to determine the point in time the problem during the scenario server or network.
8, Downloader Component Size (KB) ( Downloaded Component Size)
"Downloaded Component Size" chart for each downloaded Web pages have been set up in size. It can be seen directly by the assembly which is relatively large and need to be further optimized to improve performance.