Apstra AOS™ 2.0, the latest version of our groundbreaking product for managing networks based on intent is generally available today. I’d like to provide some further information on what is included in this release.
We announced the industry’s first vendor-agnostic Intent-Based Networking (IBN) system over a year ago enabling customers to use AOS to design, build, deploy and operate very complex Layer-3 IP fabrics in their data centers. AOS provides incredibly high levels of availability with flexibility around vendor selection to achieve lower capital and operational costs. These L3 networks are perfect for modern apps that support methods for service advertisement and reachability.
Traditional L3 Leaf/Spine design
Customers also need help to automate the creation and management of virtual Layer-2 networks to support legacy workloads. VXLANs can provide these services and it has been possible for many years, yet few people actually deploy VXLANs due to configuration complexity at scale.
Automation is a great tool to deal with complex systems, and AOS has a unique capability to abstract away these roadblocks. The highly scalable data store at the heart of our distributed operating system permits us to create simple models of what we want the network to do. AOS takes care of all of the heavy lifting in order to make it work. We are not positioning a new protocol or proprietary switch feature. Rather, AOS simply configures the network as an expert would, and then immediately begins to monitor devices to ensure that intent is being met.
AOS Data Store as the source of truth for network state
To ease the challenges of data center provisioning, we have built a very simple section in our web UI that allows an operator to create and modify virtual networks, leaving AOS to handle the tasks and validation. As always, everything in the system can be managed via a fully documented API, allowing you to integrate with a 3rd-party system like vSphere, Openstack, Hadoop, or others.
Actual AOS user interface showing five server racks and the addition of hosts to a VXLAN.
Traditional compute resources that were allocated for a specific line of business may often sit underutilized, while another department may be struggling to keep up with a huge spike in application utilization. This is a very frequent problem for CTOs and business leaders who have a hard time explaining why the network services for two environments are different. The long standing challenge of getting networks “plumbed” to various areas of the data center can be solved through the use of VXLANs, allowing for a higher return on assets. For example, the various servers in one rack may be isolated at Layer-3, which would prevent a new node from being added to an existing system due to physical and logical network constraints. VXLAN solves this problem with an industry standard encapsulation method.
Servers in separate racks that were previously isolated
are now connected over a virtual network.
Many IT leaders know that encapsulation adds a new set of challenges related to limited correlation between the overlay and underlay networks. To solve these problems, AOS contains the following technologies to help bridge these disparate logical systems:
- Physical path tracking with overlay traffic flow analysis
- Application specific tooling and dashboards (ex. Multicast)
- Live network state query engine (GraphQL)
- Real-time device location services
- Unified interface for deploying and managing physical and logical networks
- Sophisticated visualization platforms
- And more…
AOS 2.0 helps modern enterprises deal with their complex private and hybrid cloud environments. AOS has an unparalleled set of capabilities that customers are just beginning to explore and we are looking forward to working with more interested groups to help them solve the problems that have plagued their systems. Apstra has demonstrated an ability to lower both capital and operational costs while speeding up time to market for IT services.