Synthetics Overview

The Synthetics section of the Kentik V4 portal is covered in the following topics:

Performance Dashboard provides a high-level overview of your organization's Synthetics testing.
 

About Synthetics

Kentik Synthetics enables you to continuously test network performance so you can uncover and correct issues before they impact customer experience. This proactive testing is enabled by Kentik software agents (see About Synthetics Agents). The public agents (available to all Kentik customers) that make up our Kentik Global Agent Network are located in all major Internet hubs and cloud regions (AWS, GCP, Azure, IBM, etc.). You can also deploy private agents (available only to your organization) in your physical infrastructure or your cloud resources.

Ping and traceroute tests performed continuously with public and/or private agents generate key metrics (latency, jitter, and loss) that are evaluated for network health and performance. Additional types of testing available includes performance tests focused on HTTP, DNS, and full page load, as well as monitoring of BGP announcements and withdrawals.

The results from Kentik Synthetics tests are displayed graphically in the portal, both in brief on the Synthetics tab of Network Explorer and in greater detail on the tabs of the Performance Dashboard. From there you can click through either to test details for each tested link in the Test Control Center or to visualizations in Network Explorer of real network traffic over the same paths. Additionally, some metrics derived from synthetic monitoring will be displayed in other areas of the portal, such as Kentik Map, and integrated into Insights & Alerting. The metrics from synthetic tests are in many cases correlated with metrics from real traffic data that comes from Kentik's flow and SNMP monitoring.

 

Synthetic Test Types

Kentik's Synthetic Monitoring tests, which are managed in the Test Control Center in the Kentik portal (Synthetics » Test Control Center), fall into following main categories:

  • Network Meshes & Grids: With agent-to-agent performance meshes, you specify a set of agents to monitor for availability and network performance, and Kentik will automatically configure all of the required to and from testing between the individual agents. Test results will be presented as a matrix. Test types in this category currently include:
    - Network Mesh: Bidirectional tests between agents in sites within your organization's infrastructure (private agents) and/or in Kentik's worldwide network of Global agents (hosted and in public cloud).
    - Network Grid: Ping and trace tests from selected private or global agents towards targeted IP addresses, which may be entered in a comma-separated list or added in bulk from a.txt or.csv file.
    - DNS Server Grid: Test a hostname lookup simultaneously on multiple user-specified DNS servers, showing the resulting IP addresses and the resolution times, which enables you to identify servers that are non-performant.
Mesh tests monitor performance to and from a set of agents.
  • Network & Routing Tests: Network performance monitoring from a set of one or more agents that you choose toward any IP address, hostname, SaaS service, or individual agent. Kentik will automatically configure all of the required testing based on your from and to settings. Test types include:
    - IP Address: From one or more chosen agents toward an individual IP address.
    - Hostname: From one or more chosen agents toward an individual hostname.
    - SaaS Application: From one or more chosen agents toward an individual SaaS application.
    - Agent-to-Agent: From one or more chosen agents to an individual agent.
    - BGP Monitor: Based on user-specified sets of tested prefixes and "allowed" ASNs, check whether any prefix is originating from a non-allowed AS, which would possibly indicate route hijacking. If so then the route announcement that includes the hijack will be shown as a critical event in the Incident Log and as an Invalid Origin event (indicated with a red bar in the timeline) on the Test Details Page for the test.
  • Web Tests: Tests from specified agents to a specified URL:
    - HTTP(S) or API: Tests a specified Web server's response to an HTTP GET request, showing the status code, the average time to last byte, and the response size (which enables you to spot anomalies in the amount of data returned). You can also optionally run ping tests towards the resolved IP address.
    - Page Load: Tests a full browser page load using Headless Chromium run by Kentik app agents, showing the status code and the times for performance indicators response, navigation, domain lookup, etc.
    - Transaction: Tests a "transaction," which is a series of actions, running in Headless Chromium, that are driven by a Google Puppeteer script created in the Recorder tab of Chrome Developer Tools. Results include the health status and total transaction time for each agent you test from, and show any screenshots that you specified as actions in the script.
  • Autonomous Tests: Performance monitoring that is guided by the actual traffic patterns on your network, which Kentik is able to do by analyzing your flow records in KDE along with correlated network traffic data (SNMP, BGP, GeoIP, etc.). You choose which of the below test types you'd like to use, and choose the individual instance that will be the subject of the test, for example the ASN for an ASN test. We then analyze traffic and SNMP patterns to or from your network that involve the test subject, and based on that analysis we intelligently select the IPs to test from or toward (typically the IPs your traffic is using most), so your test credits will be applied most efficiently. The following test types are currently available for autonomous testing:
    - ASN: Between one or more agents and a set of autonomously selected IPs in the specified AS.
    - CDN: Between one or more agents and a set of autonomously selected IPs associated with the specified CDN (see CDN Attribution Dimensions).
    - Country: Between one or more agents and a set of autonomously selected IPs in the specified Country.
    - Region: Between one or more agents and a set of autonomously selected IPs in the specified Region.
    - City: Between one or more agents and a set of autonomously selected IPs in the specified City.
  • DNS Tests:
    - Server Monitor: Tests the performance of one or more DNS servers associated with a hostname, showing the resolution time and the resulting IP address.
    - Domain trace: Not currently implemented.
    - DNSSEC: Not currently implemented.

Notes:
- For information on configuring the above-listed tests, see Test Settings Page.
- The Synthetics module also reports BGP announcements and withdrawals (see BGP Route Viewer Tab) which, while not based on synthetic testing, enables you to correlate issues revealed by synthetic testing at the network (ping, trace) and web layers (HTTP, page load) with issues at the routing layer (BGP).

Supported Testing Configurations

Ping and traceroute testing is available for configurations such as the following:

  • From any Kentik global agents to external SaaS, application, and API endpoints.
  • From any private agents to external SaaS, application, and API endpoints.
  • From site (data center, remote office, campus, etc.) to site using private agents.
  • Between sites (private agents) and public cloud (global agents).
  • Between public cloud region instances using private agents.
 

Tests and Subtests

Kentik's Synthetics capabilities are designed to make it as fast and easy as possible to configure testing: you tell us at a high level what you are trying to test and then we largely handle the details of configuration automatically. With that in mind, we use the following terminology:

  • Subtest: An individual recurring ping/traceroute combination in a single direction, meaning from one agent toward one other network entity such as an agent, IP address, or ASN.
  • Test: A collection of subtests that are configured, run, and displayed as a single unit.

Note: The consumption of credits for a given test is calculated based on a count of that test's subtests (see Synthetic Test Credits).

 

Synthetic Test Credits

The test credits used for synthetic testing in Kentik are covered in the following topics:

Consumption of credits by your tests is indicated for both current and following month (projected).
 
top  |  section

About Test Credits

Synthetic testing in Kentik is enabled by "test credits," which are purchased from Kentik in SyntheticPaks. The purchased credits add to your organization's test credit balance. Every time a test is run this balance is reduced by an amount that is calculated based on the factors explained in Test Credit Consumption. The results of these calculations are continually updated and displayed in the portal UI (see Test Credit Display) so you always know how many credits you have to work with and what the impact of any added testing would be on your test credit balance.

Adding Test Credits

If your organization's test credit balance drops below zero a notification will appear on some of the pages in the Synthetics section of the Kentik portal. Your currently running tests will not be paused or deleted, and your organization will not be automatically charged. You will, however, be unable to resume paused tests, add new tests, or edit any existing tests in a way that increases credit consumption. To remove these limitations, click the submit a request link in the notification, which will open the Add Credits dialog, whose fields will be populated with the information that your Kentik Customer Support team will need to process your request.

 
top  |  section

Test Credit Consumption

While the per-credit pricing of a SyntheticPak may vary depending on factors such as your Kentik platform (Pro or Premiere) and the number of credits in the SyntheticPak, the consumption of test credits in your organization varies by the type of test, as covered in the topics below.

Note: The topics below apply to currently supported Synthetics tests. As additional testing types are added in the future they may consume test credits in different ways and amounts.

Network Tests Consumption

Test credit consumption by a given network test (ping and traceroute) is always based on the following factors:

  • The number of subtests (see Tests and Subtests).
  • The type of the agent from which each subtest originates (see About Synthetics Agents):
    - Global: 5 credits per subtest (one direction).
    - Private: 2.5 credits per subtest (one direction).
  • The testing interval (see Test Frequency in General Options).
  • The testing duration. Once turned on, tests typically run continuously, so in a full 30-day month a test will run 43,200 times (60 minutes × 24 hours × 30 days).

Web Tests Consumption

Credits consumption in Web tests — Page Load tests and HTTP(S) or API tests — involve a timeout factor in which the timeout in seconds (see HTTP Timeout in General Options) is multiplied by 0.5 for private agents or 1 for global agents to arrive at the credits per subtest. With the default timeout of 5 seconds, for example, the credits for a private agent subtest would be 2.5 (5 × 0.5), which is equivalent to the "network" tests described above.

BGP Monitor Consumption

The BGP monitor test is not run using agents and its frequency is fixed at one minute. Kentik charges 0.53 credits for every minute that the test is configured and running.

 
top  |  section

Credit Consumption Scenarios

To better understand how testing impacts your organization's test credit balance, let's apply the factors described in Test Credit Consumption to a couple of example scenarios. The first example is a simple mesh test (see Synthetic Test Types) between four private agents within your network. In this mesh, the four "from" agents would each test toward the other three target agents, for a total of 12 unidirectional subtests. The following table shows the test credit consumption over one month for those subtests.

Source agent Target Credits/subtest Subtests/test Credits/test Tests/month Credits/month
Private Private 2.5 12 30 -- --
Private Global 2.5 0 0 -- --
Global Private 5 0 0 -- --
Global Global 5 0 0 -- --
Total:     12 30 43,200 1,296,000

The calculation gets more complicated when considering tests involving different kinds of source agents. In the table below, for example, we look at a test involving a mesh of five agents, two private and three global, which results in a total of 20 unidirectional subtests.

Source agent Target Credits/subtest Subtests/test Credits/test Tests/month Credits/month
Private Private 2.5 2 5 -- --
Private Global 2.5 6 15 -- --
Global Private 5 6 30 -- --
Global Global 5 6 30 -- --
Total:     20 80 43,200 3,456,000

Note: The above calculations are for illustration only. In actual practice all calculations are performed by the portal automatically and results are displayed as described in Test Credit Display.

 
top  |  section

Test Credit Display

Your organization's test credit balance is displayed in various places in the portal Synthetics section, as is the impact on your credit balance of consumption by individual tests:

  • Licenses page: (Licenses): The Synthetics pane shows credits used so far this month as a percent of available credits. It also shows a count of credits used and credits available.
  • Performance Dashboard (Performance Dashboard): The Tests pane in the right sidebar shows counts of credits available, credits used so far this month, and total credits used last month.
  • Test Control Center page (Test Control Center Page): Cards across the top show current and projected Credit utilization. The % Monthly Credits column of the Tests List gives consumption this month as a percent of your organization’s total available credits.
  • Test Settings page (Test Settings Page): The expected monthly credits chart shows the credits the test will consume per month.
 

Synthetic Test Status

The status assigned to synthetic tests is covered in the following topics:

 
top  |  section

Test Status Levels

The results from every synthetic test are evaluated in three areas (latency, packet loss, and jitter) on a continuous basis to assign the test one of the following statuses for each of those areas:

  • Healthy: No issues have been found that would significantly impact performance.
  • Warning: Issues have been found that could significantly impact performance if not addressed.
  • Critical: Issues have been found that are significantly impacting performance and need immediate attention.
  • Pending: The test exists but no results are available yet.

Note: For more about how test data is evaluated for status, see Latency and Jitter Status and Packet Loss Status.

 
top  |  section

Status Time and Scope

Each test defined in the Test Control Center runs at the standard (non user-selectable) testing frequency of one minute. This testing frequency is independent of the time range (lookback duration) for which test results may be viewed, which may range from last 5 minutes to last 30 days. The longer the time range, the wider the time over which data from every-minute testing will be aggregated to create a set of "time slice" values. The following examples illustrate how this works for various time ranges:

  • Last 1 Hour: The time slice is one minute, so the latency and jitter values for the current one-minute slice will be compared against the average of the corresponding values for the 15 preceding minutes.
  • Last 6 hours: The time slice is 5 minutes, so the latency and jitter values for the current 5-minute slice will be compared against the average of the corresponding values for the 15 preceding 5-minute slices (last 75 minutes).
  • Last 1 day: The time slice is 1 hour, so the latency and jitter values for the current 1-hour slice will be compared against the average of the corresponding values for the 15 preceding 1-hour slices (last 15 hours).

Note: Time range may be set on a Test Details Page for an individual test. On the Test Control Center page the time range is always calculated as described for Last 1 Hour above.

 
top  |  section

Latency and Jitter Status

To calculate a status for latency and jitter, Kentik compares the value in the current time slice to a baseline value representing the average for the prior 15 time slices. Status is indicated as Healthy unless one of the following applies:

  • Warning: The current value exceeds the baseline by a Kentik-calculated amount, and the difference is between 1.5 and 3 standard deviations.
  • Critical: The current value exceeds the baseline by a Kentik-calculated amount, and the difference exceeds 3 standard deviations.

Note: Kentik uses a sliding scale to calculate the amounts above so that a large change in terms of standard deviation doesn't trigger a status of Warning or Critical when the amount is not large enough to realistically impact performance.

 
top  |  section

Packet Loss Status

For packet loss, the status is characterized as Healthy unless one of the following applies:

  • Warning: The current packet loss value is between 0 and 50 percent.
  • Critical: The current packet loss value is greater than 50 percent.
 
top  |  section

Overall Status

In many places within our synthetics UI the status values for latency, packet loss, and jitter are presented individually; in other places an overall status is presented. The overall status represents the worst status across the three areas. In other words, if the status of latency and packet loss are Healthy but the status of jitter is Warning then the overall status will be shown as Warning.

 

Synthetics Modules

The Synthetics section of the portal includes the following modules/workflows:

  • Performance Dashboard: The Performance Dashboard provides at-a-glance visibility into the health, performance, and availability of links between the locations (data centers, clouds, etc.) where ksynth agents are deployed as well as the IP destinations, SaaS applications, and HTTP/ DNS servers used or accessed by your network. The information is provided in a tab for your own organization’s tests, detailing incidents that occurred in a user-specified time range, and in tabs for free Kentik-run tests that tell you the current status of popular SaaS services and cloud providers. For your organization's tests there is also information on the agents deployed and the consumption of test credits, as well as the ability to drill down into information and status for individual tests.
    Note: The Performance Dashboard also reports on the following additional types of performance:
    - BGP announcements and withdrawals (see BGP Route Viewer Tab) which, while not based on synthetic testing, enables you to correlate issues revealed by synthetic testing at the network (ping, trace) and web layers (HTTP, page load) with issues at the routing layer (BGP).
    - DNS health status and resolution time from a set of global agents toward a set of commonly used DNS servers (see DNS Services Tab).
  • Test Control Center: The Test Control Center is used to manage your organization's synthetic tests. Cards across the top of the page show your current (month to date) and projected (full month, at current trends) consumption of test credits (purchased from Kentik in SyntheticPaks). A Tests List, filterable by type and by agent, provides information about your currently configured tests and enables you to drill down into the details of subtest results. You can also access an edit page for any existing test or use the Add Test button to go to the Add Test page, where you can choose and configure a new test.
  • Agent Management: The Agent Management page lets you see and manage all currently deployed agents, global (including app agents and public cloud agents) and private (see About Synthetics Agents). The location of agents is shown on a map, below which is a filterable table providing information about each agent (including status) and allowing you to activate an agent, edit its configuration, or remove it. You can also add private agents on this page using the Deploy Private Agent pane across the top.

Note: Kentik users whose level (role) is Member can view the above pages, but only Administrators (including Super Administrators) can add/edit synthetic tests or agents.

© 2014- Kentik
In this article:
×