Performance Metrics of APM Insight
Interpret APM Insight monitoring results
Monitoring your applications using Site24x7 APM Insight, an application performance monitoring tool, allows you to track and measure important metrics including apdex score, appserver throughput, response time, exceptions and more from a customizable and unified console. On the right corner you can also decide on the time frame for which you need the metrics. Options to edit, unmanage instance, delete instance, copy web script and export to PDF are also given on the top right hand side of the console. Various parameters and metrics can be obtained by accessing the following tabs:
- Overview
- Web transaction
- Database
- Background
- Traces
- Trace details tab
- Node VM (For Node.js agent)
- JVM (For Java agent)
- IIS (For .NET agent)
- Exceptions
- RUM analytics tab
- Milestone tab
- Server metrics tab
- Outages tab
- Data collection stats tab
- Instance split up tab
Overview:
The basic function of this tab is to give a bird's eye view of all the major parameters from an application's point of view.
Also, you can view details about the fastest and slowest transactions by response time as well as the recent exception traces generated.
The size of request and response objects are tracked and shown in data throughput. Request size is captured as bytes in and response size is captured as bytes out.
Data throughput helps you assess the size of the incoming request - gives you an idea as how much data your app server is handling.
For instance, from the above image you can understand that the particular transaction 'arh/trace' has a higher request size in comparison with other requests.
This hels you to assess the general size of the incoming requests and this in turn comes handy when there is a sudden spike in the size of the incoming requests.
Example, in case of DDOS attacks, the incoming requests could be of huge data size and by knowing your average request sizes, you can easily spot anomalies.
Events Timeline widget records all the events of your selected Application/Instance for a selected time range. You can identify/decode various events from the past, which includes Down, Critical, Trouble, Maintenance, Anomaly, or Suspended. Each event is color-coded for easy identification. Events can be drilled down to extract maximum data and facilitate easy troubleshooting.
Web transaction:
The various transactions running in the application are listed here along with the following metrics:
Parameters | Description |
Transaction | Name of the transaction |
Apdex | A numerical measure of user satisfaction, with 1 being the highest and 0 the lowest |
Count | Number of times a particular transaction has been called by a user |
Errors (%) | The percentage of errors that occur in a particular transaction |
Avg. Resp. Time | The average amount of time a particular transaction takes to respond to a user request |
Min | The lowest amount of time taken by a particular transaction to respond to a user request |
Max | The highest amount of time taken by a particular transaction to respond to a user request |
Total | Total time taken by a particular transaction to respond to all user requests |
Avg. CPU Time | The average amount of time taken by the CPU to respond |
Fatal | Number of fatal errors that have occurred in a particular transaction |
You can sort the available data based on values of each parameters. Also, if you select any specific transaction you will get a graphical view displaying different metrics in the form of graphs.
Database:
This section provides a complete description on the total number of SQL queries that are executed for the application.
Parameters | Description |
Database Operation | The name of the db operation being performed on the database |
Count | The number of times the db operation has been performed by users |
Errors (%) | The percentage of errors that has occurred in the db operation |
Avg. resp. Time | The average amount of time a particular db operation takes to complete |
Min | The lowest amount of time taken by a db operation to complete |
Max | The highest amount of time taken by a db operation to complete |
Total | Total time taken by a db operation to complete |
Background:
Information on the various background tasks such as maintenance, schedulers, messaging etc is provided here.
Parameters | Description |
Transaction | Name of the background transaction |
Count | Number of times a particular background transaction has been executed |
Errors (%) | The percentage of errors that occur in a particular background transaction |
Avg. Resp. Time | The average amount of time a particular background transaction takes to respond |
Min | The lowest amount of time taken by a particular background transaction |
Max | The highest amount of time taken by a particular background transaction |
Total | The total time taken by a particular background transaction |
Avg. CPU Time | The average amount of time taken by the CPU to respond |
Traces:
The various traces that have been run on various applications and all details pertaining to them are listed in this section.
Parameters | Description |
Start Time | Detailed start date and time when the trace was launched |
Transaction | The name of the transaction on which the trace was launched |
Resp. Time | The time taken by the trace to complete its execution |
Avg. Resp. Time | The average amount of time taken by the trace when run multiple times |
CPU Time | The amount of time taken by the CPU to respond |
On selecting any particular trace we can dive into individual metrics for each. The drill down is divided into 3 tabs:
- Summary
- Trace Details
- SQL Statements
- Remote calls
- JVM Metrics
- Server Metrics
Traces - Summary:
Gives an overall summary on the trace selected and the slowest components that it was able to identify.
Parameters | Description |
Slowest components | The names of all the components traced with the slowest being shown first |
Count | Number of times the component was called |
Duration | The time duration the component took to execute out of the total time that the trace took to execute |
Percentage(%) | Percentage value of the duration the component took to execute |
Custom Parameter | The user defined parameters for the respective transactions are shown |
Traces - Trace Details:
This tab allows you to dig deeper into all of the entities involved in the trace to identify the anomalous spans causing latency.
Header
This displays the total duration of the trace and the total number of spans in the trace. You can perform a quick search based on any of the trace types like APPCODE, MYSQL, WEBREQUEST, HANDLED_EXCEPTIONS, and ALL, by selecting it from the Filter drop-down menu.
Minimap
Gives a condensed view of the trace timeline. You can click and drag your mouse over the map to filter the spans of that time range. The filtered spans will be listed in the main timeline. If you want to select a different time range, click Reset and then select again.
Timeline
Shows the list of spans within the trace. You can also expand or collapse the span to view the children spans.
The screenshot below shows the expanded span.
- The timeline bar is color-coded based on the span type.
- By default, all spans are expanded except those with a time duration less than 30% of the total trace duration.
- The spans containing exceptions are highlighted in red.
Traces - SQL Statements:
Information on all the SQL queries executed by the trace
Parameters | Description |
Timestamp(second) | The time when the sql query was executed by the trace |
Execution time(ms) | How much time the query took on its own to complete |
Query | Name of the query executed |
No. of Queries | Total number of queries executed |
Traces - Remote calls:
Lists down all the external remote calls made during the execution of the trace. All the external calls made are identified and listed down under two classifications, and the total count is obtained. The two different types of external calls identified are:
- WEBREQUEST
- WCF
Traces - JVM Metrics:
For Java applications, you can navigate to Traces > JVM Metrics to get the graph views of the important JVM metrics such as JVM CPU Usage, JVM Classes Count, Heap Memory, and Non-Heap Memory before and after the trace start time. The red mark on the graph indicates the trace's start time. The user can obtain troubleshooting information by comparing the key metrics before and after the tracing.
Traces - Server Metrics:
Navigate to Traces > Server Metrics to see graph views of the server's important metrics before and after the trace start time. The red mark on the graph indicates the trace's start time. This helps the user to get information on the status of the corresponding server at the specified time.
Traces - NodeVM's metrics:
For Node.js applications, you can navigate to Traces > NodeVM's Metrics to get the graph views of CPU metrics, garbage collection data, and event loop data, before and after the trace start time. The red mark on the graph indicates the trace's start time. The user can obtain troubleshooting information by comparing the key metrics before and after the tracing.
Filter and advanced filter:
Site24x7 uses Filter and Advanced Filter options to identify traces based on multiple search conditions. You can then perform the required action on the filtered traces. You can apply the filters to any of these three categories:
All: View all traces, including error traces and distributed traces.
Errors: View the error traces.
Distributed: View the traces that flow to other applications.
Filter
You can perform a quick search based on any of the trace metrics by selecting it from the Filter drop-down and specifying the threshold value in the provided is above box.
Example: If you choose Response Time from the Filter drop-down and set the threshold value as three seconds in the is above box, all traces with response times greater than three seconds will be displayed.
Advanced filter
The Advanced Filter option has built-in AND conditions, allowing you to specify an unlimited number of search conditions. During the filtration process, the AND conditions will be implemented, meaning only traces concurring to all of these conditions will be displayed.
You can click on the Add Filters option to select the required field type—be it Transaction Name, Exception Class, Component Name, or another filter—and the field value. The field values will be listed based on the field type selected.
You can add multiple filters one by one. Each filter added will be considered as an AND condition.
For example, in the image below, the search criteria has three conditions—Transaction Name as zylker/settings/, Exception as java.lang.NullPointerException, and Response Time as above 2 seconds.
You can click the icon to view the performance of that particular transaction.
You can view the whole picture of the required trace by clicking directly on it.
Un-instrumented block of code:
In general, the APM Insight agent captures known frameworks and methods in your applications. Components involved in a transaction, including it’s method calls and functions are listed under the Traces tab.
While inspecting a transaction trace, you may encounter a field called un-instrumented block of code.
You may get this message under two circumstances:
A. When you have used custom methods or functions in your application code.
B. Even in known frameworks, the agent may not be able to track all methods or functions, called between two instrumented methods or functions. In such cases, that particular method or function is marked as un-instrumented block of code. This helps you to identify the exact occurrence of the specific method. By knowing the time stamp and the instance of occurrence, you can deploy custom instrumentation to figure out the issue.
JVM(Only available for Java agent users):
JVM tab is available only if you are using Java agent. This tab will give us all the necessary data on the JVM instance. Various parameters and metrics can be obtained by accessing the following tabs:
- Summary
- Garbage Collector
- Threads
- Configuration
JVM - Summary:
Gives an overall summary on the JVM instance with information on CPU usage, runtime memory, JIT, classes, heap and non-heap memory etc.
JVM - Garbage collector:
Detailed information on how runtime memory is being managed in the JVM instance. It lists down all the different garbage collectors that are executed along with the collected objects count and time spent
JVM - Threads:
Data on all the different paths followed in executing transactions in the application.
JVM - Configuration:
Complete description of the JVM instance running and other related information such as:
- Host name and version
- Memory type and how much is available
- Up Time
- Thread Count etc.
Node VM
The Node VM tab is available only if you are using the Node.js agent. The Node.js agent uses Node VM, a native node add-on, that collects key metrics data from the Google Chrome V8. The agent collects CPU metrics, which are usually analyzed with garbage collection (GC) metrics. These metrics help you in improving the performance of your application.
IIS (Only available for .NET agent users):
This tab is available only if you have enabled APM Insight in IIS monitor console and only if you are using a .NET agent. To know more about how to enable APM Insight in IIS monitoring, please go through our blog on the same topic. Various parameters and metrics can be obtained by accessing the following tabs:
- Summary
- Application Pools
IIS - Summary:
Gives an overall view on the IIS server and the application accessing the server.
IIS - Application pools:
Information on all the application pools that are running on the IIS server.
Exceptions:
Description of all the exceptions that have occurred with count of how many times they have taken place.
RUM analytics:
An interface tab which shows all the important data collected by Site24x7 APM Insight RUM agent.
Milestone tab:
In general, you can mark milestones to review your application performance before and after a feature update, issue fix or performance enhancements etc. You can view all such created milestones under the milestone tab. Milestones created for the corresponding application, for the chosen time period is listed here.
Clicking on a particular milestone displays your application metrics before and after the selected time period.
The following metrics can be compared using milestone markers:
- Apdex score
- App Server Response Time
- Request Throughput
- Data Throughput
- Error Count
- Exception Count
- HTTP Error Rate
- Web, Background and Database transactions
For example, clicking on the 'APM TEST' milestone, displays the application performance for a selected period, before and after one hour.
Image displaying performance of transactions before and after six hours.
You can also compare and view the performance metrics directly by clicking on the milestone view.
To view this,
- Go to Milestone View
- Click on the Milestone name
- Select the time range and click Apply.
Server Metrics tab:
In general, you can see a comprehensive list of all the server monitors mapped to the instances of your application. You can also view the server metrics at the instance level and application level separately.
Application-level metrics
If you choose the application name from the top-left menu, you will get the complete list of the server monitors associated with your application.
You can view the whole picture of performance metrics by clicking on the server monitor.
Instance-level metrics
If you choose the instance name from the top-left menu, you will get a thorough picture of all the major performance metrics of the associated server.
Outages tab:
Gives an overall summary on the Down, Trouble, Critical history of the selected Application/Instance with information on Start Time to End Time, Duration, Reason, Comments.
Parameters | Description |
Start Time to End Time | The start and end time of the detected outage |
Duration | The time duration of the detected outage |
Reason | The reason for the detected outage for quick troubleshooting |
Comments | Comments added by the user for reference |
You can click the icon to mark an outage as maintenance, edit comments, or even delete any outages if it is irrelevant.
Mark as Maintenance
You can mark a specific outage period as maintenance using the Mark as Maintenance option. After marking an outage as maintenance, the status icon changes to maintenance and the record is still available under the Outages tab. The maintenance can be reverted to outage if needed.
Data Collection Stats
You can get a detailed data report for a specific outage by clicking the hamburger icon and selecting the Data Collection Stats option.
Edit Comments
During any detected outage, Site24x7 will auto-populate the reason for the outage under the Reason section. Anyone, irrespective of the user role, can edit/delete these system-generated comments using the Edit Comments option.
Delete
You can delete any outage or maintenance that is irrelevant using the Delete option.
- You can also add an outage manually if needed using the Add Outage button.
- You can use the Download CSV button to export the displayed Outage report.
Data Collection Stats Tab:
The Data Collection Stats tab gives a detailed data report of the selected application or instance for the chosen time period. The data report will be available for the last 30 days.
Refresh the page manually to view the data report for the most recent polls.
Parameters | Description |
Status | The status of the monitor, like Up, Down, Trouble, or Critical |
Apdex | A metric for user satisfaction, with 1 being the highest and 0 being the lowest |
Average Response Time (ms) | The average time it takes to respond to a user request |
Count | The number of requests with successful responses |
Error Count | The total number of errors that occurred |
Fatal Exception Count | The total number of fatal exceptions that occurred |
Throughput (rpm) | The number of requests received per minute |
Error Rate (%) | The percentage of errors that occurred |
JVM CPU usage (%) | The percentage of CPU usage by the Java virtual machine (JVM) |
Heap Memory Usage (%) | The percentage of heap memory used by the JVM |
GC Count | The number of global garbage collections that occurred |
GC Time (ms) | The time taken to perform garbage collection |
Instance Split up tab:
In general, you can see a comprehensive list of all the instances in your application, along with the metrics associated with them.
Parameters | Description | Available For |
Apdex Score | A numerical measure of user satisfaction, with 1 representing the highest and 0 representing the lowest. | Java, .NET, Node.js, PHP, Ruby, Python |
Satisfied (count) | The number of transactions that are labeled as Satisfied. If any transaction response time scores values below the Apdex threshold value, the transaction is labeled as Satisfied.
|
Java, .NET, Node.js, PHP, Ruby, Python |
Tolerating (count) |
The number of transactions that are labeled as Tolerating. If any transaction response time is exactly equal to Apdex threshold, or in between satisfied and frustrated threshold values, it is labeled as Tolerating.
|
Java, .NET, Node.js, PHP, Ruby, Python |
Frustrated (count) |
The number of transactions that are labeled as Frustrated. If any transaction response time scores above four times the Apdex threshold, the transaction is labeled as Frustrated.
|
Java, .NET, Node.js, PHP, Ruby, Python |
Resp.Time (ms) | The average time taken by the instance to respond to user requests. | Java, .NET, Node.js, PHP, Ruby, Python |
Throughput (rpm) | The number of requests received per minute. | Java, .NET, Node.js, PHP, Ruby, Python |
Req.Count | The total number of requests received. | Java, .NET, Node.js, PHP, Ruby, Python |
Errors (%) | The percentage of errors that occurred. | Java, .NET, Node.js, PHP, Ruby, Python |
Status | The status of the instance, like Up, Down, Trouble, or Critical. | Java, .NET, Node.js, PHP, Ruby, Python |
Host | The hostname of the instance. | .NET |
IP |
The IP addresses of the instance. Hovering over the value displays the complete IP list.
|
Java, .NET |