Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

Step 2: Up and Running

 

Look Around the Juniper Mist Portal

The Juniper Mist platform you’re setting up includes Wired Assurance Service for automated operations and service levels, Health Statistics for detailed visibility into EX Series switches, and a plethora of device and network details that you can drill down to from most of the dashboard pages.

The image below illustrates how Juniper Mist cloud services use telemetry from the physical infrastructure to provide you with AI-driven health insights through the Juniper Mist cloud.

Our Virtual Network Assistant, which we call Marvis, uses natural language processing (NLP) for troubleshooting and applies continuous learning through supervised machine learning self-driving capabilities. Marvis Actions come in both self-driving and driver assist modes. As you’ll see, you can use Marvis to list wired clients connected to the network and troubleshoot any corresponding issues that it identifies (such as speed mismatches, missing VLANs, switch health, and anomaly detection). Marvis helps identify the root cause of issues across various IT domains (WLAN, LAN, WAN, and security), and automatically resolves issues within its purview. In addition you can use the Juniper Mist portal to do the following:

  • Streamline troubleshooting (SLEs, root cause, anomaly detection, Marvis Actions).

  • Automatically apply a Junos OS configuration setting to new switches based on common templates that cover most enterprise features. For corner cases, you can always access the switch and use its CLI from the portal.

  • Enable dynamic port configurations to work with any RADIUS server based on user group, MAC OUI, or LLDP name match.

  • Set up flexible port assignments with manual or dynamic configurations, using port profiles and templates that are based on connected endpoint type.

  • Measures pre-connection and post-connection metrics with SLEs.

OK. Let’s see how all this works.

Network Visibility

The main Monitor page in the Juniper Mist portal shows an overview of the network. From here, you can you drill down into specific events and connected devices, see the pre-connection and post-connection experience on the wired network, and identify switches that are seeing anomalies on the same site.

  1. Log in to your organization, or to the public Live Demo (if you have an account). The Live Demo, which we use in these screen shots, has a variety of connected devices and live analytics that you can explore.

    Here’s what the Monitor page looks like:

  2. From the Monitor page, scroll down and start exploring on your own. The depth and breadth of what you can do from this page is too much to describe here, so take some time to click around. Use the online help to learn about whatever interests you.
  3. Let’s look at an example based on another view of the network, this time, Wi-Fi coverage mapped in the context of a building floor plan. Click Switches in the menu, and then select the Location tab to show the distribution of switches across the floor plan of an office building.

    In this view, you can see all the switches and access points, where they are situated, and the respective health status of each. Furthermore, you can click any device to show its status, including which devices are connected to it.

  4. To see whether there are any Wi-Fi dead spots, click 2.4 GHz and the 5 GHz in the Wi-Fi Coverage box on the right side of the page. You’ll see visual representation of the coverage. This makes it easy to understand users' Wi-Fi experience in the area.
  5. Click any device name on the map to drill-down to its configuration and health details. When you do, an AP Details link becomes available in the Summary box that appears about half-way down the page.

Here’s an example showing 2.4 GHz Wi-Fi Coverage when it’s mapped across a floor plan:

Service-Level Expectations

Now let’s see how you can track service-level expectations (SLEs) for wired, wireless, and WAN connections using the dashboards. Juniper Mist lets you set service level thresholds for all the major things that impact wireless performance, including time to connect, coverage, capacity, roaming, and network uptime. If any of your parameters are violated, you can have the system alert you about the reasons why the parameter is not being met, the top mobile devices affected, the top wireless networks affected, and so on. For example, you might have a goal of 2 seconds for the minimum time it takes mobile users to connect to the Wi-Fi. With one quick look, you can see that this SLE is only being met 72 percent of the time. Something is wrong. Let’s find out what.

Here’s what the Monitor page looks like after drilling down to Wireless SLEs:

  1. Click Monitor in the menu, then Service Levels, and select the Wireless tab.
  2. Since we’re interested in Time to Connect, click that to drill down and see the Juniper Mist insights about what factors are causing the delays.

    Here’s what the Monitor page looks like after drilling down for a root cause analysis of the poor time to connect statistic:

  3. As you can see in the break out classifiers that are listed alongside the Time to Connect service level metric, the cause is clearly “Association.” This is responsible for 100 percent of the delays. Without getting too far into the weeds, let’s just say the Association statistic calculates the time it takes from the start of the mobile client’s association packet to the time when the client successfully moves data. Anything longer than two seconds is flagged.

    In the chart at the bottom of the page, you can jump to the Correlation chart to see which devices are implicated—or better yet, let’s just jump to the Summary chart to see what Mavis has to say about it:

    If you were curious about the deeper cause, from this same page, you could click Networks > WLAN and then Live_temo_SLOW_demo to dig into the configuration settings of that particular WLAN, and resolving the issue.

Switch Health

Now let’s take a quick look at the Switch Health page, where you can get a root-cause analysis of any of the following service level metrics: throughput, successful connections, and switch health.

  1. Click Monitor in the menu on the left, and then Service Levels. Choose the Wired tab at the top of the page and then click the chart that says Switch Health.

    Here’s what the Switch Health page looks like after drilling down for a root cause analysis of the poor Switch Health statistic:

    As you can see, Switch Health is at 30 percent, and a quick glance at the break-out classifiers shows the reason why: CPU usage is at 97 percent, which means the CPU is implicated in 97 percent of the issues associated with this switch.

  2. In the chart that appears below the CPU breakout details, click Distribution to identify the switch by name, and then, if you want, click the name to drill further down and see what is going on.

Access Points

Juniper Mist access points provide a Wi-Fi connection to users (and devices) within the target area, using a wired connection to an EX Series switch. Juniper Mist access points are also purpose-built to collect metadata for over 150 states that the cloud-hosted AI engine uses to perform analytics, machine learning, location service, and event correlation. They incorporate a third radio for always-on security monitoring, troubleshooting, and synthetic client tests.

In the Juniper Mist portal, you can view all the access points in a given network or site from the Access Points page. From there, you can quickly assess the number of clients attached, the devices capabilities (such as Wi-Fi or Bluetooth or both), device uptime, and other details.

Let’s explore this page now.

  1. Click Access Points in the menu on the left to open the page.
    • If you have any unassigned access points in the network, they’ll appear at the bottom of the page. You can claim them by clicking the Inventory button at the top of the page and then the Claim APs button that appears in the upper-right corner of the resulting page.

    • If you want to troubleshoot access point issues, select an access point from the list and click the AP Insights link in the Access Point details page. The Access Point Insights page appears.

    Here’s what the Access Points Overview page looks like:

Marvis Actions

Bad cables can be a nasty bit of business in the networking world. Cables rarely ever fail, so when they do, no one ever thinks to look for a bad cable as the root cause of the ensuing issues. Marvis can detect bad cables easily by inferring the cause from the network symptoms that surround it. Note that on your newly deployed switch, chances are you won’t have any issues to find right now. The same is true for the Live Demo site, which might or might not currently have this exact issue to show. Nevertheless, you can see the issue illustrated in the Marvis Actions dashboard, and that is what the following steps describe. By the way, bad cables are just one of the many, many things Marvis detects—we’re looking at it only because it provides a quick and clear illustration of Marvis’ scope.

  1. Click Marvis in the menu on the left, and then click the Actions button in the upper right-hand corner of the page that appears.

    Here’s what the Marvis page looks like after drilling down into a Bad Cable:

  2. “Switch” appears in the chart with a 3 next it, indicating that Marvis has detected three issues with the device. To drill down and see what they are, click Switch and expose the three issues. One of them is “Bad Cable.”
  3. Scroll down to see the Marvis Actions report, which says the following:

    Recommended Action

From this same page, we can see that Marvis is also reporting a Negotiation Mismatch, and an issue detected in one of the access points connected to the switch. Upon further examination, the issue turns out to be an “EAP/802.1X Failure”, which means the RADIUS server cannot be reached. Marvis identifies the probable cause, and suggests the solution as follows: These APs are missing as NAS client on the Radius server. Please add them in order to resolve the issue.

The analytics truly are rich.