Networks are often seen as a bottleneck due to legacy box-by-box complex workflows that includes hop-by-hop troubleshooting, high resolution time for issues and manual change management operations. Traditional norms of managing data centers isn’t enough anymore. Network teams managing on-premise data centers are burdened by operations that are slow, complex and are unable to keep up with 24x7 uptime requirement. In order to offer application teams speed, agility, reliability, and experience of hosting applications on public clouds,  there is a need to re-design on-prem data centers with the same design principles and techniques used internally by cloud providers.

Big Switch’s Big Cloud Fabric (BCF) provides public cloud-style networking experience for on-premise enterprise data centers . BCF simplifies lifecycle management of workloads (hosts, VMs, containers etc),  starting from bootstrapping up to application workload provisioning and all the way up to ensuring business continuity. BCF also offers consistent tool chain to manage workload connectivity and security policy on private cloud, hyperconverged environments, public cloud and provides richer endpoint visibility.

In this blog, we will focus on how BCF offers cloud-inspired Zero Touch Fabric (ZTF) capability for Day0/Day1/Day2 operations, so that customers truly experience an “invisible” network - that is, a network that operates at the speed of VMs and containers.

 

 

Consider a case where you have applications or service running in a virtualized environment on VMware SDDC suite. In order for your datacenter to be ready for virtualization, the underlay (physical) network has to be provisioned first. Switches and servers need to be mounted and cabled appropriately, followed by installing software and booting-up the gear. Network admins are then required to individually configure and provision switches to function as a leaf-spine fabric. This underlay provisioning process thus becomes time consuming, cumbersome and labor intensive. 

“Time is really the only capital that any human being has, and the only thing he can’t afford to lose.”
—Thomas Edison

Challenges with traditional on-prem operations: 

  • Host network provisioning: Manually provision underlay fabric and specific ports on switches with appropriate configurations for LAGs/LACP, etc
  • Transport provisioning: Define and trunk transport VLANs on multiple switches, increasing the number of touch points and possibility of misconfigurations.
  • Visibility: Manually correlate underlay failures from different consoles which requires log scraping from individual switches.
  • Troubleshooting: Manual box-by-box analysis required to uncover underlay connectivity issues, to isolate switch or data-path causing the traffic drops.

Modernizing data center network operations: 

How does Big Switch Networks' Big Cloud Fabric modernize data center operations?

Big Cloud Fabric is inspired by cloud networking design principles. It provides automated provisioning, centralized management and real-time monitoring of leaf-spine based pod fabric running open networking switches through BCF Controller software.

Optimized Day0 operations with BCF: 

In order to add a switch in your fabric, mount the switch on a rack and cable it appropriately.  On the BCF Controller dashboard, enter switch name and select MAC address from an auto-discovered MAC address list. Once switch name and MAC information is added to the controller, it generates initial configuration for the switch. Next, the switch downloads the NOS software and configuration from the BCF Controller and initiates the boot-up process. BCF Controller automatically forms leaf/spine fabric from the connected switches. Similarly, when a switch is RMA’ed and replaced by a new switch in the fabric, it will automatically inherit the configuration from the BCF Controller and would not require any manual provisioning by the network admin.

 


Optimized Day1/Day2 operations with BCF: 

BCF Controller orchestrates life-cycle management for the entire fabric including software upgrades, snapshots etc thus eliminates the need of logging into the switch. BCFs’ architecture ensures hitless upgrade of the entire fabric and enables admins to perform on-demand or scheduled software upgrades from BCF Controller in a matter of minutes. Recently a manufacturing customer in midwest upgraded an 80-switch BCF fabric (2 controllers, 4 spine switches, 76 leaf switches) in ~20 minutes without impacting application traffic.

BCF simplifies network and host provisioning and ensures underlay network is always in parity with the virtualized workloads. BCF brings public cloud constructs like VPC/VNet constructs on-prem and offers this for enterprises as Enterprise Virtual Private Cloud (E-VPC). E-VPC can be delegated out to the tenant, or self configurable, and or can be uniquely assigned  to that tenant via RBAC control. BCF offers rapid troubleshooting with its fabric trace functionality that allows the network admins to view path between application end-points, across the leaf-spine-leaf fabric and isolate error condition causing traffic drops.

In summary, BCF simplifies Day0/Day1/Day2 operations for the data center fabric as follows: 

  • One-click switch addition/removal workflow
  • Automated fabric discovery and leaf/spine formation
  • Automated fabric wide software upgrade/installation via BCF Controller
  • Never login to individual switches for anything

Contextual Visibility and Analytics:

BCF provides real-time context specific insights to enable network admins to mitigate the issue as soon as it is detected. BCF enables detailed fabric-wide visibility for logical and physical components: 

  • Physical topology with connected switches (links etc)
  • Real-time hardware details like front panel port configuration and status
  • Power supply, fan configuration and status. Visibility into server CPU and memory usage.
  • Connected devices page shows all the physically connected servers 
  • Endpoints and VPCs visibilityTest-path for application endpoints 

 

BCF provides contextual VM, container and host level visibility through integration with on-prem cloud platforms (VMware SDDC, Nutanix, Red Hat and Container). It offers granular insights for difficult to diagnose network condition like mis-configured policy, missing routes, unresolved ARPs etc. The comprehensive built-in analytics provides proactive insights into the fabric for anomaly detection, for instance, it identifies at risk interface which has higher risk of dropping packets.

 


BCF also has 3rd party integration which blends very well in DevOps environment, for instance, Ansible-based automation and Slack notifications generated for customized error/alert conditions.

Next Play: Zero-touch Hybrid Cloud Networking:

BCF, with its Zero Touch Fabric capability simplifies underlay operations and provides actionable intelligence to NetOps/DevOps/CloudOps teams for rapid root cause analysis & troubleshooting. With our BCF  journey, we are focused to bring VPC networking experience everywhere and bridge the gap between on-prem and cloud networking thus empowering enterprise datacenter fabric deployments to operate at speed of VMs and containers.

Important Links:


Thanks,
Anshu Verma
Senior Product Manager - Big Switch Networks