Automating Scale-Out IP Fabrics with the Adaptive Cloud Fabric

SHARE:

New L2 Overlay Features Simplify Private and Hybrid Clouds

Pluribus customers have been building and automating scalable data center network infrastructure for years using the uniquely powerful architecture of the Pluribus Adaptive Cloud Fabric, the only open network solution that automates modern “scale-out” IP fabrics and incorporates overlay network virtualization to create flexible Layer 2 (L2) and Layer 3 (L3) services across the fabric. Our newest release makes it even more powerful, adding a rich new set of Layer 2 overlay features that simplify building private and hybrid clouds for a wide range of applications.

Before discussing the new L2 capabilities, let’s take a step back and review how to build a scale-out IP fabric and why it’s so important to incorporate network automation using software-defined networking (SDN) and network virtualization.

What Is a Scale-Out IP Fabric?

A “scale-out” data center network architecture, also called horizontal scaling, is a modern approach to building data center network capacity that has several benefits compared to traditional vertical, or “scale-up,” architectures.

A scale-out infrastructure is based on linking together a number of relatively small, fixed-form-factor network switches that collectively provide the required capacity. Because new capacity is added by simply adding switches, a scale-out architecture provides a predictable operational model where both network capacity and cost grow linearly with the number of network elements, a huge advantage over the traditional vertical approach.

The industry’s best practice for building scale-out architectures is to create a fabric based on IP (i.e., Layer 3) with a multi-stage Clos topology, which offers a non-blocking and resilient connectivity model with predictable performance and scaling characteristics.  Traffic workloads are efficiently distributed across the network using equal cost multipath (ECMP) load balancing.

Figure 1 shows an example of an IP fabric in a single data center “pod.”

Figure 1: Data center pod example with 16 leaf switches providing 768x 10G ports with 1.2:1 oversubscription

While scale-out IP fabric architectures are great for scalability, resilience and performance, they can be operationally complex to provision, monitor and troubleshoot. What’s more, L3 fabrics do not natively support L2 connectivity requirements.

How Does Network Automation with the Adaptive Cloud Fabric Simplify Scale-Out IP Fabrics?

The Adaptive Cloud Fabric dramatically simplifies scale-out IP fabric operations by combining a number of innovations. A key foundation is its controllerless SDN architecture, with a unique distributed control plane that enables comprehensive network automation and simplified management of both the IP fabric itself (which forms an “underlay” network) and virtualized network overlays.

All fabric nodes can be managed from any single point, whether they are colocated or geographically distributed, while guaranteeing consistent configuration across nodes. The result is a dramatic reduction in configuration complexity, fewer opportunities for human error and easier integration into higher-layer orchestration systems to achieve a fully automated software-defined data center (SDDC).

Other key technology attributes of the Adaptive Cloud Fabric that enable comprehensive, simplified network automation for scale-out IP fabrics include:

  • Network virtualization based on highly available VXLAN overlay fabrics with automatic provisioning, enabling flexible L2 and L3 connectivity services that are decoupled from the physical underlay network topology
  • Scalable overlay technologies, including Anycast Gateways, that enable workload and application mobility across multi-site data centers without complex routing or IP addressing changes
  • Scalable endpoint database with conversational forwarding, enabling full network visibility and control from any network device
  • North-to-south fabric interconnection using distributed virtual routing function (VRF) technology

What Are the Adaptive Cloud Fabric’s New L2 Capabilities?

Private and hybrid cloud architectures often require transparent Layer 2 Ethernet transport, both within private clouds (across single or multi-site data centers) and for handoff to public clouds, all while maintaining multi-tenant isolation via network segmentation. Transporting L2 traffic transparently across an L3 fabric requires some form of overlay network virtualization, and in multi-tenant cloud environments, it’s also critical to expand traffic tagging options beyond basic VLAN tagging, with its restriction to 4,094 unique VLANs.

The Adaptive Cloud Fabric has enabled transparent L2 service overlay connections for some time with service constructs such as Virtual Link Extension. Now, by combining VLAN stacking technology (informally known as “Q-in-Q”) with new VXLAN-based bridge domain objects, the Adaptive Cloud Fabric can scale up to 16 million services for up to 4,000 customers, with flexible traffic hand-offs to external cloud service providers and support for a wide variety of connectivity models and service types, as shown in Figure 2.

Figure 2: Hierarchical fabric structure using bridge domains for intra- and inter-DC connectivity

The result is an unprecedented new level of L2 flexibility for building private and hybrid clouds.

The Adaptive Cloud Fabric was already the best way to build a scale-out IP fabric, and with these new L2 capabilities, it just keeps getting better.

Where Do I Learn More?

To learn more about all of the above mentioned technology innovations, check out our updated Technical Brief: Achieving a Scale-Out IP Fabric with the Adaptive Cloud Fabric Architecture.

SHARE:

SHARE:

Subscribe to our updates and be the first to hear about the latest blog posts, product announcements, thought leadership and other news and information from Pluribus Networks.

Subscribe to Updates

SHARE:


About the Author

Jay Gill

Jay Gill

Jay Gill is Senior Director of Marketing at Pluribus Networks, responsible for product marketing and open networking thought leadership. Prior to Pluribus, he guided product marketing for optical networking at Infinera, and held a variety of positions at Cisco focused on growing the company’s service provider business. Earlier in his career, Jay worked in engineering and product development at several service providers including both incumbents and startups. Jay holds a BSEE and MSEE from Stanford and an MBA from UCLA Anderson.