Monitoring application load balancer metrics

Metrics available by service plan

The supported monitoring metrics include:

• Active connections to your application load balancer at a given time
• Throughput of data passing through your load balancer over a given time
• Connection rate, or an analysis of when more or less connections are made to your load balancer
• Total number of HTTP/HTTPS requests received by the back-end
• Average response time for HTTP/HTTPS requests
• Count of responses with the HTTP 2xx back-end response code (successful request)
• Count of responses with the HTTP 3xx back-end response code (redirect)
• Count of responses with the HTTP 4xx back-end response code (client side failure)
• Count of responses with the HTTP 5xx back-end response code (server side error)
• Count of total back-end members in a pool
• Count of active back-end members in a pool

These metrics help track the traffic and usage patterns for your load balancers and can provide insight about peak traffic hours, usage drop offs, and overall usage patterns.

Each metric is composed of the following metadata types:

• Metric name - The name for the collected metric.
• Metric type - Metric type determines whether the metric value is a counter metric or a gauge metric. Each of these metrics is of the type gauge, which represents a single numerical value that can arbitrarily fluctuate over time.
• Value type - A unit of measurement for a specific metric. Examples include bytes or counts. A value type of none means that the metric value represents individual occurrences of that metric type.
• Segment - How you want to divide and display the monitoring metrics.

Only the resources in the same zone as the selected subnet can connect through this VPN gateway.

Metric segmentation

You can split the data into various visualizations in the IBM Cloud Monitoring dashboard, allowing views of different metrics based on your preferences. For example, if you have multiple load balancers or accounts with different load balancers in each account, you might want to focus on a particular listener port.

As an example, you can segment the active connections by Application load balancer listener port to show how many active users are connected to the load balancer through each listener type. To illustrate this, let's assume that your load balancer has two different listener protocols one HTTP on port 80 and another for TCP on port 8080. The dashboard would contain different lines showing 10 users who are connected through HTTP on Port 80 in one color, and 6 users connected through TCP on port 8080 in another color.

Enabling metrics monitoring

To receive monitoring metrics, you must set up IBM Cloud Monitoring with a monitoring instance.

To do so, follow these steps:

1. Navigate to the metrics monitoring portal, then click Create a monitoring instance.

2. Select a region for your monitoring instance.

   If you do not have an existing load balancer, see Creating an IBM Cloud Application Load Balancer for VPC to provision one.

   The region should match the location of your existing load balancer.

3. Choose your pricing plan.

   Pricing plan details are explained in the selection window. Select the plan that best meets your requirements.

4. Provide a service name for your instance. It can be any name that you want, and has no impact on functionality.

   Do not create multiple monitoring instances with the same name.

5. Optionally, select a resource group. A resource group is a way to organize account resources in customizable groupings. Any account resource that is managed by using IBM Cloud Identity and Access Management (IAM) access control belongs to a resource group within your account.

   If you do not have any pre-configured resource groups, or no reason to share this resource selectively, use the default selection.

   If your account has multiple resource groups, you can choose which one has access to this monitoring instance. This allows you to have metrics available to some resource groups and not to others.

6. Select the Enable Platform Metrics checkbox. You must select this to receive metrics from your load balancer.

7. Click Create. You are taken back to the monitoring metrics home page.

Within a few minutes, your new monitoring instance is displayed with several configurations. You might have to refresh your browser to see it.

Working with the IBM Cloud Monitoring dashboard

To view and work with your metrics, follow these steps:

1. Navigate to the metrics monitoring portal.

2. Click View IBM Cloud Monitoring next to the service name of the monitoring instance that you want to work with.

   The first time that you access your monitoring instance, several windows display as part of the internal setup. Leave these selections with their default entries, and click through the pages until you reach the main page.

3. Select Dashboards on the left sidebar to open the IBM Load Balancer Monitoring Metrics dashboard. Then, click Default Dashboards > IBM > Load Balancer Monitoring Metrics. The default dashboard is not editable.

4. Eleven main metrics in the dashboard are shown: Throughput, Active Connections, Connection Rate, Request Count, Request Latency, HTTP_2xx Response Count, HTTP_3xx Response Count, HTTP_4xx Response Count, HTTP_5xx Response Count, Total Members and Active Members. You must create a custom dashboard to modify parameters and segment of your metrics by load balancer ID or listener port.

   You can choose what time window you'd like to see your metrics displayed for, using the bar at the end.

Creating a custom metrics dashboard

You can create your own dashboard to customize your monitoring metrics, such as viewing information about particular load balancers, or seeing traffic that comes only through particular listeners.

To customize your dashboard, follow these steps:

1. Navigate to the metrics monitoring portal.

2. Click View IBM Cloud Monitoring next to the service name of the monitoring instance you want to work with. The dashboard displays.

3. On the left sidebar, select Dashboards. Then, click the green + sign in the panel.

4. Select Blank dashboard, then select the type of visual representation you want.

   IBM Cloud Monitoring offers eight different visualizations for your dashboard. Read the description for each visualization, then choose the one that best meets your requirements.

   Line ("View trends over time") is the easiest and most basic option. It is also the most frequently selected option. The examples in this topic show a Line-based visualization.

5. Configure your custom dashboard.

   • In the Metrics field, enter ibm_is to display the three load balancer metrics: ibm_is_load_balancer_active_connections, ibm_is_load_balancer_connection_rate, and ibm_is_load_balancer_throughput.

   You can monitor listener port traffic by enabling the ibm_is_load_balancer_listener_port metric.

   • You can choose a scope to display in your dashboard by clicking Override Dashboard Scope. For example, you can display the metrics for a particular load balancer.

   • You can also set a segment to compare metrics across the scope that you defined. For example, you can look at throughput for a particular load balancer segmented by listener port.

6. Click Save for your new custom dashboard to be accessible.

   By default, the dashboard begins with the name "blank dashboard". You can change the name by selecting Dashboards from the sidebar, then clicking the Pencil icon next to the name.

To return to the default dashboard at any time, select Dashboards > Default Dashboards > IBM > Load Balancer Monitoring Metrics.

Working with IBM Cloud Monitoring using the APIs

You can also work with the monitoring instance by using the metric query APIs. You might want to do this if you need raw data points or want to consume your metrics from a command-line interface rather than using the IBM Cloud Monitoring dashboard.

After creating your monitoring instance, you must collect the following two pieces of information.

• The Monitor API token
• The endpoint of your IBM Cloud Monitoring instance

To collect this information and work with your monitoring instance using metric query API, follow these steps:

1. Access the Monitoring home page, and click Open Dashboard next to the instance you want to work with. After the dashboard displays, select your Account Profile icon on the left sidebar, then select Settings. Your account settings display.

2. Your Monitor API token is an alphanumeric string that is located in the Monitor API Token field. Click the Copy button to the right of the key to transfer it to your clipboard.

   Do not share this API token. Anyone who has this API token has full access to your metrics.

3. To get the endpoint of your IBM Cloud Monitoring instance, navigate to your main dashboard in your browser. Then, select the URL to the dashboard, which appears similar to:

   https://us-south.monitoring.cloud.ibm.com/#/default-dashboard/ibm_is_load_balancer?last=3600
   

   The first part of the URL (in this case, us-south.monitoring.cloud.ibm.com) is your endpoint. Make note of it.

4. After you have both the API token and the endpoint, you can format your POST request. The following POST request is an example, with all the parameters that you can modify. These parameters are:

   • The Monitor API token.
   • The endpoint of your monitoring instance.
   • The value for ibm_resource (this is the load balancer ID you want to see metrics for).

   If you want to see this metric for all of your load balancers, do not enter a value for the scope attribute. For example, use "scope" : "".

   • The metric type that you want to see the results for. This example uses ibm_is_load_balancer_throughput, but ibm_is_load_balancer_active_connections and ibm_is_load_balancer_connection_rate are also valid options.
   • The from and to attributes define the times to focus the scan, set in Epoch Time and in microseconds.
   • The sampling and value attributes set the granularity of which data is returned in the POST request.

Because a large volume of data is stored in IBM Cloud Monitoring, choosing the specific level of granularity is important. IBM Cloud Monitoring can return only 600 data points at any time with a given request. As a result, the sampling and value attributes are important. Leaving these two lines out of your request returns an aggregate sum over that time period instead.

If the time range specified by from and to is large (for example, 4 days), but you define a sampling and value of 10 seconds, this means that you receive 4 days worth of data that is split into 10-second chunks. This is not a useful sampling due to the large amount of data returned. Specifying a larger chunk is recommended (for example, 1 hour instead of 10 seconds).

curl \
-H 'Authorization: Bearer <API_TOKEN>’ \
-H 'Content-Type: application/json' \
https://us-south.monitoring.cloud.ibm.com/api/data/batch  \
-d '{
 "requests": [
     {
         "format": {
             "type": "data"
         },
         "scope": "ibm_resource = \"cfc851b1-f30b-4a06-b354-5b64526c0a69\"",
         "metrics": {
             "k0": "timestamp",
             “v1”: "ibm_is_load_balancer_throughput"
         },
         "time": {
             "from": 1584396900000000,
             "to": 1584402600000000,
              “sampling”: 600000000
         },
         "group": {
             "by": [
                 {
                     "metric": "k0",
                     “value” : 600000000
                 }
             ],
             "aggregations": {
                 “v1”: "sum"
             },
             "groupAggregations": {
                 “v1”: "sum"
             }
         }
     }
 ]
}'