Cisco today introduced Application-Centric Infrastructure as the vision for Next Generation Data Center architecture, built for both today’s physical and virtual workloads as well as tomorrow’s highly dynamic Cloud-based, and performance-intensive big data application environments. Please check out Padmasree Warrior’s blog or Cisco Unified Fabric to learn more.
What I would like to share with you is how we are evolving the Cisco Unified Fabric to deliver operational simplicity through superior integration.
Introducing Cisco Dynamic Fabric Automation (DFA)
Delivering Operational Simplicity through Superior Integration
As organizations accelerate private and public cloud deployments, IT organizations and data center networks must evolve to meet rapidly changing and growing requirements. Virtualized and cloud environments require more agility and simplicity to quickly deploy and migrate virtual machines. IT organizations, on the other hand, are challenged with operational complexity, architectural rigidity and infrastructure inefficiency with manual processes, disjointed provisioning, deficient software overlays, static resource allocations and disruptions when growth is needed.
The good news is that Cisco continues to evolve its Unified Fabric to address these needs. The new Cisco Dynamic Fabric Automation delivers unsurpassed operational simplicities through superior integration. It does this by …...automating device configuration…automating fabric configuration for virtual machine (VM) deployment and migration…integrating bare-metal and virtualized resources without the need for a dedicated gateway…providing tenant awareness to the fabric management layer…simplifying day-to-day operations…and supporting scalable multi-tenant cloud deployments.
The new Cisco Dynamic Fabric Automation delivers:
1- Optimized Fabric Infrastructure for Enhanced Efficiency and Scale
With enhanced forwarding, distributed control plane and integrated physical and virtual, Cisco DFA is optimized for spine-leaf topologies and simplifies networking by enabling any network anywhere with seamless mobility for physical and virtual machines.
Leaf switches function as the Layer 2/3 boundary with an integrated gateway and connect all resources, including VMs, bare-metal servers, physical and virtual services, and WAN access. Spine switches provide high-speed connectivity and forwarding between the leaf nodes.
This topology provides high bisectional bandwidth, deterministic 2-hop latency between resources. It also delivers greater resiliency with smaller failure domains and multi-tenant scale of greater than 10,000 tenants/networks.
Cisco Dynamic Fabric Automation also lets VMs communicate with bare-metal servers or physical services without requiring a dedicated gateway. This significantly improves performance and scale by eliminating gateway bottlenecks and points of failure.
2- Simplified Fabric Management with Open APIs for Ease of Operations
Cisco Prime Data Center Network Manager (DCNM) 7.0 provides centralized fabric management allowing you to manage the fabric in a holistic manner. It enables automated network provisioning, common point of fabric access, and host, network and tenant visibility.
Cisco DFA simplifies switch deployment significantly. With Power-On Auto Provisioning, when a switch is added to the fabric, Cisco DCNM identifies the switch as a leaf or spine, and automatically delivers the correct configuration files and switches are up and running in minutes.
Cisco DCNM also simplifies operations by providing a central point of management across your physical and virtual networks…from fabric topology discovery…to mapping tenants to switches…to fabric monitoring and diagnostics of every network device in the fabric.
With open northbound REST-based APIs, Cisco DCNM better integrates with Cisco and 3rd party orchestration and automation tools, in addition to cloud platforms.
3- Automated Provisioning for Greater Agility
Cisco DFA enables network automation and provisioning for simplifying both physical servers and virtual machine deployments and migration across the fabric. Based on the network admin defined network profile templates for physical and virtual machines, instances of Network Policies are automatically created in DCNM when a Server Admin provisions VMs/PMs. When a VM/PM pertaining to a project is detected, Network Policy is applied to the network leaf.
Here’s how it works.
When a new tenant is provisioned, the tenant network information is sent to the DCNM and the Virtual Supervisor Module (VSM). When a VM is instantiated on a server, the VSM passes the port profile information to the Cisco Nexus 1000V. The Nexus 1000V then forwards the tenant ID to the leaf switch using VDP, an industry-standard protocol. The leaf switch then queries the DCNM to retrieve the tenant network information, and applies it to the port.
And that’s it!
A new tenant and a VM have been provisioned without any manual network configuration, significantly reducing deployment time. Of course while this works better with Nexus 1000V, Cisco DFA also supports other vswitches.
For VM mobility, reprovisioning the fabric is just as simple. When a VM moves from one leaf switch to another, the destination leaf switch follows the same process, automatically downloading the VM’s tenant information and configuring the port.
Introducing Cisco Nexus® 7700 Switches
Cisco is extending the Nexus 7000 Series with the addition of the Nexus 7700 switches. These new Nexus 7700 switches offer industry’s highest scalability in an environmentally optimized form factor while maintaining operational, feature and architectural consistency across the entire Nexus 7000 Series. Nexus 7700 delivers:
- Up to 83 terabits per second (Tbps) of overall switching capacity
- Industry’s highest-capacity 10, 40 and 100 Gigabit Ethernet ports – with up to 768 native 10-Gbps ports, 384 40-Gbps ports and up to 192 100-Gbps ports.
- The Cisco Nexus 7700 Switches deliver a comprehensive set of features with nonstop operations in two chassis form factors: Nexus 7718 and Nexus 7710
What is really important to note is that the Nexus 7000 and the Nexus 7700 leverage the same fundamental building blocks, such as identical ASICs and software releases. This offers our customers a consistent operational model between the platforms.
Introducing F3 Series Modules
Cisco is also announcing new 40G/100G optimized F3 Series modules for the Nexus 7000 switches and Nexus 7700 switches. Based on Cisco’s custom F3 ASIC, and integrated with Cisco’s NX-OS software, the F3 modules deliver the industry’s broadest data center switching feature set in a single module offering.
- For the Nexus 7000 switches: a 12 port 40G module and a 6 port 100G module
- For the Nexus 7700 switches: a 24 port 40G module and a 12 port 100G module
- Some of the features include FabricPath, VXLAN, Fabric Extender technology (FEX), FCoE, OTV, LISP, MPLS along with Cisco DFA.
This comprehensive feature set simplifies customer deployments by giving our customers the flexibility to deploy a wide range of network designs with a single module, ensuring unprecedented investment protection and agility to adapt to changing networking requirements over time.
In addition, the F3 Series modules represent a 60% improvement in power efficiency, helping to address the environmental challenges in today’s data centers.
BOTTOM LINE IS THIS…
Some of the world’s largest data centers run on Cisco Unified Fabric – Nexus and MDS portfolios. And there is a reason for that! And now with the latest Cisco Dynamic Fabric Automation innovations, your IT organization becomes even more agile, cost effective and efficient. Unlike other vendors in the industry, Cisco Unified Fabric enables your data center to be more simplified, optimized and automated, significantly simplifying VM deployment and migration and overall data center operations. This in return, frees up your IT staff to concentrate on taking care of initiatives that really matter to the business and increasing your bottom line!
To learn more about Cisco Unified Fabric, Nexus and MDS Switches, please visit Unified Fabric.
Will the F3 modules support bidirectional PIM?
Can you outline where the M-series vs F-series differentiators lie now that the new F3 cards have greater L3 functionality?
It is true that the F3 series integrate a lot of the M Series functionality. However, the M Series modules still have some unique value in terms of larger table sizes (FIB, ACL, MAC), Full Netflow support, and MacSec on all ports. Further, there are some interface types only supported on M series (Native 1G Copper/Fiber, X2 optics, etc).
As for BiDir PIM, we currently support this on the M series and are investigating whether we will add software support on the F series. The F3 does have hardware support for BiDir.
Hope this helps.
Need to know whether there is any limitation when we mix the F1/F2 and M1/M2 cards along with the new F3 cards in an existing N7K chassis?
Our current plan of record for F3 is to provide full Layer2 and Layer3 interoperability with M2 as well as with F2/F2e already at FCS.
Comments are closed.