Data Center

Building Highly Scalable SDDC Infrastructures with VMware NSX and Brocade VDX

by Anuj Dewangan ‎08-24-2015 11:30 AM - edited ‎01-28-2016 01:32 PM (12,612 Views)

As IT organizations evolve into service providers for internal customers, the supporting data center infrastructures require a rich set of highly scalable network services with high resiliency. As a result, the core requirements for these data centers now resemble that of public cloud service providers. In fact, many private clouds are now being built on the same guiding principles as large data centers. It is imperative for data center architects to design and build infrastructures that can scale while reliably providing the required services and meeting the tenant service level agreements (SLAs). 


During our upcoming VMworld session, we will look at reference architectures and best practices for building the physical network as well as deploying VMware NSX for the Software Defined Data Center (SDDC), with specific focus on design principles that allow the infrastructure to scale to large deployments.


The proposed architectures for the physical networks for the SDDC are built on the following design principles:

  • Simple
  • Modular
  • Repeatable
  • Highly scalable

We will specifically review building scalable physical networks for SDDC sites with the following architectures:

  • Optimized 5-stage folded Clos architecture using L2 PoD deployments
  • 3-stage folded Clos (Leaf-spine) and optimized 5-stage folded Clos architectures using L3 PoD deployments

A snapshot of the physical network built using Brocade VCS Fabric technologies in an optimized 5-stage folded Clos architecture is shown below.


Picture 1.png

The snapshot below shows an L3 Clos network in an optimized 5-stage folded Clos architecture.




We will compare and contrast the Clos architectures and see how requirements drive the choice of deployment.


Next a review of how VMware NSX components fit in the physical infrastructure and how vCenter and NSX components are deployed in this network to ensure scalability and resiliency for the workloads and the network infrastructure will be performed. We will specifically look at:

  • NSX components and their placement in the physical network
  • Best practices for deploying multiple vCenter instances for large scale SDDC deployments
  • vCenter host and HA cluster design

The snapshot below shows HA cluster design and multiple vCenters across datacenter server racks.




We will also look at how built in automation mechanisms of Brocade VDX switches, VMware vSphere and NSX, allow infrastructure to be easily provisioned and managed leading to significant savings in operational expenses. We will specifically look at the following:

  • Auto provisioning of new network and server infrastructure
  • Network automation solutions for change management

If you are looking to build a datacenter that is highly automated and scales elastically across the physical and logical networks, as well as provide rich services like multitenancy, distributed routing and switching, micro-segmentation, VPN and load-balancing, then come join us for the session!


Note: This blog was originally written in September for VMworld US. It has been updated to reflect Brocade's presence at VMworld Barcelona in October.




Anuj Dewangan

Technical Marketing Engineer,

Brocade Communications Systems Inc.

CCIE 36868



Elver Sena Sosa


Sena IT Services

CCIE 7321