Last updated
Last updated
WAN Insights uses data from your SD-WAN to forecast the projected quality of every path on your network. Based on this forecast, WAN Insights shows you recommended path changes that will improve the quality of your end users’ experience. Here’s a quick tour to get you started.
In the left-hand navigation menu on the ThousandEyes platform, choose WAN Insights > Recommendations. The Recommendations screen appears, showing the top 4 recommendations within each application category. Each card shows recommended path changes for a single site in your network, along with improvements you can expect to see.
On the Recommendations screen there are two tabs at the top called Top and All.
Top shows the top 4 site recommendations per application category. What you’ll see initially are the out-of-the-box application categories that come with WAN Insights, such as Voice, Google Workspace, and Office 365.
For best results, look for recommendations that offer the following combination:
Both significant quality improvements, and
More users are impacted
The first thing you’ll want to do is add in any business-critical applications that you would like WAN Insights to monitor. These are applications that you want to generate recommendations for.
Note that vManage offers a number of pre-configured applications for general categories including Voice and Google Workspace. These default categories already display in WAN Insights.
On the WAN Insights Recommendations screen in the Top tab, click + Add application in the upper right. In the modal that appears, the applications should be visible under Available, assuming that there is traffic associated with them. Each “application” is actually a category or bundle of applications with similar traffic characteristics. Click the row with the plus sign, slide to Active, and click Submit.
The recommendations appear immediately on the WAN Insights Recommendations screen.
The All tab on the WAN Insights Recommendations screen shows all recommendations with Search and Filter fields up top. For example:
Search by country name to see any recommendations for sites within that country.
Show sites of Topology type Hub only.
Show all the recommendations for the Voice application category.
Show Favorites, which are categories that you’ve marked as a Favorite.
Note that if no better path exists that works overall better on a site level, you won’t see any recommendations for path changes for that site.
Next, we’ll investigate why this recommendation was generated.
Let’s find a recommendation where current path quality is low and potential improvement is high. Choose a recommendation for a site topology of Spoke for your first tour, as Hub sites are more complex.
The description below shows sample values that you might see on a recommendation card.
For the above example, here’s a brief explanation of the items on the recommendation summary card:
Application icon - Let’s say that this recommendation card is for the Voice application category, for the site shown at the bottom of the card. In this case, the site is located in the city of Bad Homburg vor der Höhe, Germany.
Use gold - The recommendation is to switch all your Voice traffic originating in Bad Homburg vor der Höhe to use the circuit identified in your SD-WAN as gold.
Current: blue - The current path is what you’re using now to handle most of your outgoing Voice traffic from Bad Homburg vor der Höhe. In this case, most of the traffic is using the circuit called blue.
(78.5%) - The number in parens (%) indicates the current path quality from the last 7 days, which is also the expected path quality of your current paths. Quality scores are derived from the three basic path quality measures of loss, latency and jitter, and are a weighted measure of risk. A quality score of 78.5% indicates that at least one of these metrics is at risk of exceeding the Service-Level Objective (SLA) thresholds defined for Voice application traffic.
Issued: 224 days ago - You can see when the recommendation was created and how long it’s been active. Recommendations are calculated based on a rolling time window of the past 7 days, and it means that the same path conditions have persisted the majority of the time since the recommendation was created 224 days ago.
Site 347, Germany > Bad Homburg vor der Höhe - This identifies the city and country of the site.
12 users impacted - Number of active users on this site, who are impacted by quality issues. The user icon indicates the number of users who are active at the site in Bad Homburg vor der Höhe, who are impacted by the lack of quality for Voice applications. Making the recommended path changes could thus improve user experience for up to 12 people.
Now that we’ve reviewed the summary information about this recommendation, it’s time to investigate why this recommendation was created. How has the recommended path performed in comparison to the current path?
Some questions to consider include:
How many connections does this site have to routers (endpoints) on other sites?
Are they all showing problems or is it just one or two?
When did problems first appear in the timeline?
For particular endpoint pairs, what is the circuit quality from one hour to the next?
For endpoints showing problems, how do these problems correlate with the loss, latency, and jitter?
How many users are impacted for each endpoint?
Click the recommendation card to drill down to the next level, which shows a 30-day quality timeline with user counts, as well as a visual breakout showing which routers have the most impacted users, and the biggest potential quality gains.
The very top of the modal repeats information from the recommendation card.
In the example above,
Recommendation shows
Projected quality of the recommended path
Projected improvement over current path quality (which is shown as Current Quality if you hover over the detail graphs below)
A Status of Ready means this recommendation is currently valid.
Below the summary information banner is a Site Quality Comparison timeline. Initially the timeline shows the last 7 days of recommended versus current quality. The past 7 days is the evaluation period that was used in the quality forecast, which has kept this recommendation active (Ready) as of today’s date.
Use the swimlane on the timeline to scroll back in time and see quality comparisons for up to 30 days. Look for time periods that show both:
High numbers of impacted users, and
Big quality differentials between recommended and current path quality
Sometimes the lines may cross. It’s possible to have a short time period where current path quality is actually better than the recommended path. However, recommendations are based on the long term projections, and overall the recommended path is still the best choice. If that changes, WAN Insights will eventually withdraw the recommendation or create a new one.
The Evaluation period refers to the rolling 7 days’ worth of historical data that is currently being used to inform this recommendation. On the timeline, the evaluation period shows as cross-hatching.
If you see gaps or if the timeline looks like it stops and starts, that means that for some hours, there was no application traffic. This could happen on a weekend or when no one is working onsite.
If you hover over a portion of the timeline, you’ll see detailed statistics. You can eventually use these hover statistics to compare the higher-level and lower-level data views as you drill down to the endpoint detail level.
These are applications that are already configured via AAR in vManage as described on this Cisco SD-WAN page.
See for information on WAN Insights permissions and ThousandEyes user roles.
See for information on managing users, roles, and permissions within your ThousandEyes organization.
+21.53% Expected quality improvement as forecasted. The + % number indicates the amount of improvement that you could expect if you took the recommended action. In this example, you could improve your Voice user experience by 21.53%.
Request to Apply - Clicking the Request to Apply button bookmarks this recommendation in vAnalytics. See for more information.
The next step is to review endpoint pairs to try to isolate areas that are having problems. This will show the root cause or why this recommendation was created. Is the problem mainly in one set of routers, or is it every router, and every connection on this site, that is affected? In the second part of this tour on , we will review endpoint pairs.
