Enterprises live in an ever-changing environment, which makes onboarding, deploying and managing devices at scale complex and time consuming. Additionally, the growth of clients, applications and IoT devices within campus and branch environments only makes matters worse, causing IT teams to struggle to meet network service level expectations within their user communities. CIOs need a way to tackle these problems head-on, while keeping user experiences front and center.
At Juniper Networks®, our Experience-First Networking foundation plays a critical role in how we deliver enhanced user experiences, with simplified and automated IT operations and insights from client-to-cloud. By harnessing the power of AIOps, driven by Mist AI™, we’re delivering real results for our customers, including top companies (Dicker Data, Wolt), educational institutions (Ashland Public Schools, Dartmouth College), hospitals (Jan Yperman Hospital, El Centro Regional Medical Center) and retail stores (Gap Inc., Outletcity Metzingen) across the globe, while earning distinction from the industry’s premier analysts along the way.
In fact, as reflected in the latest earnings announcement, Juniper’s second quarter of 2021 was a record quarter for wired access, spurred largely by unparalleled operational benefits delivered via the Juniper Mist™ cloud and Mist AI coupled with best-in-class hardware, like the recently announced EX4400 Ethernet Switch. Similarly, it was another record year for wireless access, with triple-digit growth in the Juniper Mist portfolio.
New Enhancements to the AI-Driven Enterprise
Today, we announced the next chapter in our Experience-First Networking journey, once again distancing ourselves from the competition with new enhancements to the AI-driven enterprise portfolio. These new enhancements make it even easier to deploy and operate campus and branch networks at scale:
- AI-driven campus fabric management via the Juniper Mist cloud
- New Marvis AIOps Actions that provide deeper insight for fast problem remediation
AI-Driven Campus Fabrics with the Juniper Mist Cloud
Many of today’s campus networks leverage proprietary technologies and complicated L2/L3 architectures that weren’t designed to meet modern requirements. Additionally, mismanagement of configurations and lack of scale make it difficult to maintain an agile environment with mobile users and IoT devices. EVPN-VXLAN is a common and open standard that’s often used to address these challenges, but it can place added operational complexity on IT teams to maintain configuration consistency across deployments. Juniper solves this operational burden with the Juniper Mist Cloud, allowing EVPN-VXLAN campus fabrics to be easily deployed and managed. With the Juniper Mist solution, administrators can simply choose a topology and let the software do the rest.
Using the Juniper Mist Cloud for campus fabric environments means the EVPN-VXLAN fabric is managed by the same platform that manages wired access functions in the campus – Juniper Mist Wired Assurance and Marvis, the Virtual Network Assistant (VNA). This not only provides a common operational framework for all wired access deployment, configuration and Day 2 operations, but it also delivers on AIOps, simplifying and automating operations with a single AI-driven support model. This enables self-driving operations and predictive actions that can alleviate problems before they arise.
With the Juniper Mist solution, customers can also leverage a common operational schema across LAN, WLAN and WAN environments, which is a key part of the Juniper client-to-cloud differentiation. Based on this common schema, a Global 10 energy company and a leading UK supermarket chain recently chose Juniper to handle their complete wired, wireless and SD-WAN needs.
The Juniper Mist Cloud currently supports the following devices as part of AI-driven campus fabric configuration management:
- EX2300, EX3400, EX4300, EX4400, EX4600, EX4650
- QFX5110 (-32Q, -48S), QFX5120 (-48Y, -48YM, -48T, -32C)
Marvis Conversational Interface
As Mist AI covers more use cases, it becomes increasingly important to build knowledge graphs within the system beyond what many consider the most important one – troubleshooting. In addition to Marvis’ continued growth, documentation and product knowledge, these graphs also complete the vision of a true AI conversational assistant. Marvis uses the troubleshooting knowledge graph to discover or resolve issues and can even offer recommended actions.
Ease of Use
When it comes to questions that Marvis may not yet have the answers to, the documentation knowledge graph points IT admins to detailed guides and steps to complete configuration tasks. This is especially helpful for IT admins who may not be familiar with Juniper Mist products. The Marvis Conversational Assistant guides administrators, helping them learn, configure and manage the Juniper Mist products and takes admins directly to the product or feature in question.
Extending Marvis Actions
Our vision for Marvis is that it becomes a member of the IT team, offering recommendations and determining faults based on the knowledge provided by Mist AI. Marvis Actions empowers IT staff with a set of recommendations verified by Mist AI. The power of these modern microservices cloud and rich AI toolbox is seen in the constant evolution of the Marvis service. With the introduction of several new Marvis Actions that span from client-to-cloud, Juniper is delivering great experiences for both end users and IT administrators. This includes:
- Persistently Failing Wireless Clients
- Persistently Failing Wired Clients
- ARP/DNS Anomaly Detection
- AP Bad Cable
- AP Coverage Hole Detection
- Insufficient RF Capacity
- Continuous Wired Port Flap
- Bad WAN Uplink
- Bad Cable Detection for 3rd Party Switches
Marvis Actions: Persistently Failing Wired Clients
In Figure 3 below, the Persistently Failing Wired Clients example identifies two clients that are experiencing persistent issues across a large deployment. These clients could be legitimate clients with connectivity issues causing a poor user experience or they could be rogue clients continuously adding unnecessary traffic to the network. Determining what action to take is up to the IT admin, but Marvis Actions isolates the details of the failure and speeds time to resolution.
Marvis Actions: Bad WAN Uplink
In the case of a bad WAN uplink, Marvis pinpoints the issue highlighting the need to check the connection, as shown in Figure 4 below. By identifying issues like high latency and jitter, Marvis can indicate that the link is causing a poor user experience downstream. Using details from wireless clients, APs and switches, Marvis is able to determine and isolate the root cause. That’s the power of Mist AI – with Marvis, IT administrators can now track down and mitigate these “needle in the haystack” problems with ease.
We’re excited about the latest enhancements to the Wired Assurance, Wireless Assurance and Marvis VNA software subscriptions. In addition to providing real, tangible benefits to Juniper’s customers and partners, these enhancements underscore the robustness of Juniper’s AI-driven enterprise solution and commitment to Experience-First Networking.
At Mobility Field Day 6, Juniper also announced a slew of new, complementary product enhancements, including streaming dynamic packet capture, roaming service levels, an Android client, Marvis integration with Slack and much more. Check out the presentations and demos here.