“Arista is unshackling the complexity of today's legacy multi-tiered box approach to an elegant Spline-based campus design. Once again, we are disrupting the status quo to deliver a cognitive architecture with behavior-driven actions,” said Jayshree Ullal, president and CEO for Arista Networks.
Extending the Cloud Network to Campus
Today’s campus networks suffer from too much complexity brought on by the myriad of platforms, operating systems, proprietary features and network management tools from incumbent vendors. Coupled with the explosive growth of endpoints as well as the requirement for workloads, users and devices to be connected anywhere, the operational costs of managing these complexities become prohibitive.
These challenges are not unlike those of legacy datacenters, before the shift to cloud networking. The cloud networking principles of simplification, open-standards, software-driven control, born in the datacenter, are just as applicable to today’s campus networks.
Arista’s Universal Cloud Network (UCN) delivers common cloud principles for simplified networking topologies and architectures across use-cases. As siloed Places-In-the-Network (PINs) of the legacy enterprise are normalized to become standardized Places-In-the-Cloud (PICs), Arista’s UCN provides a consistent experience and simplifies the model for enterprise customers to extend their datacenter networking practices to the campus networks.
Introducing the Arista 7300X3 and 7050X3 Campus Spline
As the first step in addressing the campus network architecture, Arista is introducing the 7300X3 and 7050X3 Spline, high performance 10/25/40/50/100G Ethernet platforms, extending the open UCN architecture from the datacenter to the campus. As Spline platforms, the X3 Series collapse multiple tiers of legacy hierarchical campus designs into a single tier. Arista Spline customers have realized simplified network designs with fewer touchpoints while still achieving high availability levels for campus voice and video. Proven in the datacenter, the Spline platforms provide a consistent set of switching and routing features including a cognitive approach to automation, segmentation and visibility.
Helping customers diagnose congestion related network performance issues, Flow Tracer, the latest feature in Arista’s telemetry portfolio, provides end-to-end visibility for any flow. The 7300X3 modular Spline system scales to 50 Tbps of performance with a choice of linecard options. The 7050X3 fixed Spline complements the 7300X3, offering a choice of compact and flexible options including 25G and 100G that deliver the same campus network consistency.
A Cornerstone: The Cognitive Management Plane
The data networking industry has seen tremendous growth in the last 40 years, aligned around a set of mature, multi-vendor data plane and control plane options. However, the management plane remains in stark contrast with no consistent or common approach. This leaves the burden to the customer, who is challenged with stitching together vendor-specific management solutions.
The Cognitive Management Plane (CMP) is open framework designed to address this gap for large data sets. The CMP combines a state repository, a stream computation engine, and various application components built into a horizontally scalable cluster. Each cluster manages a subset of network devices from different vendors and interacts with other clusters through vendor-neutral APIs and standardized models, such as OpenConfig with gRPC.
Modern streaming collects the network state to provide full network state history, where machine learning techniques are applied, providing insights including anomalous behavior detection. The goal of the CMP approach makes modern, multi-vendor network management a reality, filling the void in existing management solutions.
Securing the Campus with Cognitive CloudVision
CMP-based turbines can detect network issues that legacy systems have routinely missed and ultimately help to reduce the mean time to identify and remediate these issues. CloudVision federates the state across network types (datacenter, cloud, campus, etc.) and can then share this data with peers in the CMP framework. CloudVision can also display the data to the CloudVision front-end, visualizing both real-time and historical telemetry data with perspectives ranging from high level topology-wide views down to device level details. CloudVision couples this new level of visibility with automated provisioning, giving customers the ability to detect and take action for ongoing operational tasks.
Further, CloudVision provides compliance audit checks, with native alerting for operational situations ranging from deviations in the intended network configuration to the rollout of a security patch. Finally, Macro-Segmentation Services (MSS) leverages an open-standard approach to service insertion for the datacenter or the campus with direct firewall integration from Arista's security partners.
Now campus operators are able to leverage these operational and security tools across cognitive use-cases.