May 30, 2018

Apstra, Dell EMC, Awnix Reduce TCO in Tier 1 Service Provider

datacenter_tco_400x267I am excited to share the first production Apstra AOS® deployment of OpenSwitch (OPX) in a US-based Tier 1 telecom service provider. This deployment delivers a highly automated network operations solution we’ve developed in collaboration with Dell EMC and Awnix. To deliver on this telco service provider’s requirements for transformative TCO, agility, and reliability, the customer deployed a private-cloud based on the following technologies:


  • Apstra AOS®
  • Awnix Openstack-based ARC Cloud Platform
  • Dell Z9100-ON 100GigE Open Networking switches running the open source OpenSwitch (OPX) operating system.

Competing in the Cloud Era Requires Log-scale TCO Improvements

Apstra delivered a game-changing operational model for network services, enabling the following benefits:

  1. Powerful intent-based automation of network services: Day 0, Day 1, and Day 2+; including the physical network, as well as the VXLAN-based overlays and security zones connecting virtual servers and containers, such as Kubernetes (K8S). These services result in a massive reduction in box-per-box manual operations and manual labor.
  1. Complete real-time visibility into the network state and powerful continuous validation that the network is delivering on intent, enabling log-scale improvements in reliability, as well as a simplification of the tool chain. That is, there is no need for integration with other monitoring tools, or manual box-per-box troubleshooting.
  1. Powerful auditing capabilities enabled by AOS: a single source of truth, sophisticated RBAC, time machine like capabilities, and audit-trail — all these features deliver unmatched auditability, security, and control.

The conventional manual operation of networks is severely challenged in delivering present-day cloud services and controlling TCO in modern data centers. This manual approach cannot compete with the highly automated operation practiced in the so-called public cloud data centers. Telcos and other cloud providers, who are building data centers to deliver cloud services to their customers, are impacted. Simply put, these telecom providers gain a significant and much needed competitive edge through AOS which delivers an order of magnitude improvement in TCO, agility, and reliability.

Automation + disaggregation = massively reduced TCO

Taking a step back, open networking and open APIs — in short, disaggregation , is the reason this Tier-1 service provider selected AOS, the most powerful automation and visibility tool on the market. According to AWNIX CEO Rick Kundiger, “AOS is “the best management and monitoring tool I’ve seen for networking in decades.”

In choosing Apstra AOS, extending beyond the benefits of eliminating manual operation, they were able to realize the promise of disaggregation and take advantage of the choices that are available to them: choice of hardware, capacity, ASIC, switch operating system, and choice of the cloud platform.

In an effort to reduce TCO, this telecom provider was able to choose the latest, most cost-effective hardware that delivers on their capacity and support requirements. The solution is based on the Dell Z9100-ON switch that packs 32 x 100GigE ports in a small 1U, 195W form factor. This capability provides 200Gb/s of bandwidth to each cloud hypervisor with each Z9100-ON capable of up to 6.4Tb/s non-blocking switching I/O, ensuring plenty of bandwidth is remaining at 100s of Gb/s to Tb/s of available bandwidth for east-west and uplink traffic.

They were also able to select OpenSwitch (OPX) as their open source switch operating system,   which delivered on their specific Layer2 and Layer3 protocol capabilities and requirements, enabling the service provider to deploy OPX in production and consequently achieve a significant reduction in CapEx.

This deployment is a proof point that both automation and open source switch operating systems are mature and now available to Telcos and cloud providers, making it possible for them to compete with hyperscale companies without hiring an army of Stanford CS PhDs – an exciting milestone for the industry!

Put customers in control: the power of choice, risk-free

It is important to emphasize the role that AOS plays – beyond automation – in enabling the choice that comes with disaggregation. Through its Intent-Based approach that features powerful abstractions and network services composition, Apstra AOS re-aggregates the disaggregated components. To quote Yousef Khalidi, Corporate Vice President for Microsoft Azure Networking, “Apstra’s networking solution supports our customer needs for automation and analytics and will help bring the benefits of SONiC to enterprise data center deployments.”

Because of Apstra AOS’s uniform intent-based operational model across all supported switch OSs and hardware choices, customers can seamlessly swap between those products without having to invest in re-education, or change their operational procedures.

Last but not least, at Apstra we recognize that integration and interoperability costs are a major component of TCO. Therefore, while we’ve designed AOS to be deeply open and extensible, we are committed to only offer choices to our customers that we have fully tested and can stand behind – choices that we provide full support for as part of our turn-key solutions.

Without Apstra AOS, choice can equate to complexity as the end user is left having to integrate and test interoperability of the components themselves. Choosing a new platform or new product, such as a switch from a new vendor, incurs significant costs. Infrastructure teams are forced to shift their focus away from important strategic priorities and instead spend significant time learning the details of yet another switch operating system.

With Apstra, customers can take advantage of choice – both open alternatives, such as OPX, SONiC, and Cumulus or from established vendors such as Arista, Cisco, or Juniper – they can do so without incurring the risks or costs of interoperability, integration, and education. Customers can get the benefits of those choices, risk free to boot.

Since we mentioned established vendors, it is worth noting they are following suit and disaggregating their own offerings. In a blog post following OCP Roland Acra announced that Cisco is offering the option to buy Cisco Nexus hardware without Cisco’s NX-OS switch operating system. To quote Roland, “while these APIs obviously play a critical role for our service provider and web-scale customers, openness also means having flexibility to choose the hardware and software combinations that best meet their needs.” He went on to say, “We’ve also made it possible to run NX-OS on third-party hardware platforms – independent from our Nexus switches.  Now, customers can leverage our industry-leading software innovations and gain the flexibility to adapt them to hardware platforms that best suit their needs.” Yes, disaggregation enables customers to use AOS automation with Cisco devices – and indeed, Apstra has supported Cisco equipment since our v1.0 release.

Quadrupling performance improvements in hardware in a few short years!

Last but not least, Apstra also enables customers to seamlessly adopt the latest and greatest networking equipment through its powerful automation of Day 2 operations. With Apstra’s solution, customers can literally swap equipment in minutes – a process that typically can take weeks, if not months.

To illustrate the importance of upgrading infrastructures to the latest capacities, let’s compare the Dell Z9100-ON switch to a switch from a mere four years ago, the Dell Z9000. The Dell Z9100 delivers 32 100GigE ports in a 1U form factor. Its predecessor, the Dell Z9000 packed 32 40GigE ports in a 2U factor, and consumed 30 percent more power.

This means that the Z9100-ON offers 2.5 times more bandwidth at approximately 75 percent of the power, in nearly 50 percent the size. Or, on a per bps basis, the Z9100 is 3.25 times more power efficient, and 5 times smaller. This means it will consume less than a third the power, and take up 5 times less space in your data center! All while providing 150 percent more bandwidth to every cloud hypervisor.

With such massive savings and performance gains, organizations need to carefully evaluate whether it is better to invest in the new technology, which also enables the associated automation, than continuing full depreciation cycle with the old. As Gartner has observed, organizations may need to upgrade their infrastructures – even mid-cycle if needed, and invest in greenpatch strategies that provide them the benefit of the new – a disaggregated, fully automated infrastructure that takes advantage of the latest technologies, including Apstra AOS, to achieve log-scale improvements in TCO, agility, and reliability.

Onwards and upwards

In closing, I would like to invite other telcos, cloud providers and web-scale companies to follow the examples of Yahoo! Japan Corporation and this Tier-1 telco service provider and give Apstra a spin. We provide risk-free try-and-buy programs so you can experience the power of “unmatched feature velocity” and the “best management and monitoring tool in decades!”

Mansour Karam

President, Founder