Since it was founded in 2004, Arista Networks Inc. has been known as a company that makes great network switches. This has been done with great hardware and a single, open operating system that spans all its products.
In 2015, the company launched “CloudVision,” which moved the management, orchestration, and automation capabilities into the cloud. Historically, CloudVision has taken a backseat to Arista’s network infrastructure, which makes sense given the switches have been Arista’s core differentiator since its inception. This week, Arista announced several updates to its CloudVision platform, making it easier for businesses to manage their entire network — across data centers, campuses and cloud environments.
Typically, organizations need help managing different network parts using separate tools, technologies and operational models. For example, an information technology team might use one set of tools for managing data center operations, another for handling campus networks and another for cloud services.
This siloed approach has resulted in higher operational costs and complexity. It has made it harder to coordinate tasks, troubleshoot issues and maintain consistent security across the entire organization. On top of that, it has become more challenging to integrate, expand and upgrade the network.
CloudVision aims to eliminate the siloes by unifying network management in a single platform. Originally designed to meet the rigorous demands of cloud service providers, CloudVision now extends these benefits to enterprises of all sizes. CloudVision stands out for its ability to maintain “feature parity” with Arista’s Extensible Operating System or EOS. This means CloudVision can support all the same features as EOS without requiring constant updates.
In traditional network management systems, updates to network operating systems could result in a lag of 18 to 24 months before management tools catch up. With CloudVision, any new feature added to EOS is automatically supported so enterprises can take advantage of the latest network capabilities without delays.
CloudVision is built on Arista’s Network Data Lake, a centralized data set that collects and consolidates data from across the network. Using artificial intelligence-driven analytics and predictive insights, network operators can identify and tackle potential issues before they escalate.
“Our enterprise approach is to build the best-in-class network,” said Jeff Raymond, vice president of EOS product management and services at Arista. “Our open standards, design principles and high availability capabilities — all allow us to do that. On the operations side, this is where CloudVision comes in, we focus on simplifying and reducing the headcount needed to operate the network. Then, of course, security has become a big part of our story as well.”
Here’s a closer look at how CloudVision supports different network areas:
In data centers, CloudVision provides real-time insights and automation for easier network management. CloudVision Universal Network Observability or CV UNO uses machine learning to detect and resolve problems. CV UNO provides a view of network and application data, showing how different parts of the infrastructure interact. This enables faster troubleshooting and more efficient network management.
In campus environments, which often involve a broader range of users, CloudVision simplifies operations by offering guided workflows that cover network provisioning, management and troubleshooting. The platform addresses everything from day-zero setup to day-one and day-two operations, such as making port changes, adding or removing networks and running diagnostics.
A key feature in campus networks is the Ask AVA autonomous virtual assistant. The generative AI chatbot is integrated into CloudVision and allows operators to interact with the platform using natural language. This makes it easier for non-expert users, such as help desk staff and junior network administrators, to navigate the system.
For the wide-area networks, CloudVision provides operator dashboards a view of the WAN connections using telemetry to monitor network health. One of CloudVision’s features for WAN management is geographic mapping, which charts WAN links’ location and traffic patterns. The platform incorporates dynamic routing, using real-time data to choose the most efficient paths for application traffic. It also integrates with Google Cloud Platform and Amazon Web Services, providing a unified view of both on-premises and cloud-based WAN networks.
“The key to our approach is we have this much broader data set, especially with our cloud service offering, where it’s global and anonymized,” said Raymond. “So we can start to train data and look at a much bigger sample size. This is just the start of what we’ve built into CloudVision from a generative AI perspective.”
Lastly, CloudVision enhances zero-trust security through multidomain segmentation. It provides orchestration and monitoring services for zero-trust policies so organizations can enforce security rules based on micro-perimeters. These policies are managed through dedicated security dashboards, making it easier for network operators to visualize and control traffic flow across the enterprise. CloudVision works with firewalls like Palo Alto Networks, ServiceNow for IT management, and VMware for virtualization, giving companies a complete view of their network security.
A holistic view helps organizations simplify the management of hybrid environments and maintain consistency across the network. Many enterprises have already adopted CloudVision, starting with data centers and expanding to campus networks, WAN and security environments. CloudVision has evolved and become an integral part of Arista. When customers adopt Arista, CloudVision is automatically included as part of the overall solution rather than an optional add-on.
“We have been late to all the markets we’ve entered, and we’ve been able to build the product correctly,” Raymond said. “It ends up playing out well in the long run. That applies to EOS, our network data lake that we continue to invest in, and CloudVision at the application level.”
For Arista, it’s good to see the focus on CloudVision. Its growth has been driven primarily by its best-in-class hardware and software, with the cloud complementing the strategy. We are rapidly moving into the AI for networking era, creating a disruptive opportunity for the company. My recent AI networking survey found that a whopping 93% of customers will consider changing vendors if the AI capabilities are demonstrably better.
AI excellence will be delivered from the cloud, and while CloudVision is a strong platform, there is no clear leader in AI for networking. Raymond was clear there is more to come with CloudVision as Arista looks to leverage the cloud for future innovation.
Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.