The new unpredictability of bandwidth consumption drives automated networks
A rapidly changing data ecosystem…
As virtual reality, IoT, social media, artificial intelligence, machine learning, and other advances have grown over the last five years, applications have changed and become more diverse. This in turn has had a profound impact on bandwidth consumption and has made planning much more complicated, with applications now consuming bandwidth in a dynamic and highly unpredictable manner. As the servers that deliver these various applications continue to multiply, they have become more powerful and have higher bandwidth connectivity, requiring data center networks to provide more bandwidth.
…drives transformation of traditional network architectures
This new requirement for applications delivered on demand, coupled with the introduction of multi-tenant server technologies (VMs and software containers) and DC orchestration software, has prompted an unprecedented transformation of data center networks to make them as automated as possible for provisioning and dynamic tuning of performance. Traditional networking methods to provide users with a quality experience, such as statically assigned priorities (QoS), are no longer effective. The network must be able to recognize individual application traffic flows and adjust priority to match the dynamic nature of application traffic in a resource-constrained world.
...and the advent of software-defined networking (SDN)
The demand for automated and more efficient data center networks sets the stage for software-defined networking in the data center, which is where SDN had its initial success. At the same time, the data center network was evolving to allow efficient east-west traffic flows, creating a data center fabric. SDN control had two architectures: centralized and distributed. The distributed control architecture for SDN offered higher reliability, and, as it turns out, was well-suited for the next phase of data center SDN.
Edge computing and 5G on the horizon…
As the market continues to progress to 2020, the consolidation phase for data centers is giving way to a renewed movement to distribute compute where needed, enabled by the development of applications that use software containers and micro-services. In effect, compute is becoming natively multi-site, crossing the physical boundaries of individual data center sites and requiring SDN control across this physical boundary. Looking out to the horizon, compute will be located closer to the user with the rollout of 5G networks, meeting the needs of low-latency applications with additional evolutions of the network to support computing across hundreds or even thousands of locations.
…means growing SDN deployments
Respondents to the latest Data Center SDN Strategies Survey from IHS Markit indicated their top SDN DC use cases revolved around automating the data center network, which matches our expectation that SDN will be applied for automated provisioning and deployment of applications distributed across multiple sites. As SDN matured in 2018 deployments, respondents expected a rapid increase in production trials in 2019, then a surge in live production from 2019 to 2020.
As seen in the latest Data Center SDN Network Equipment Tracker by IHS Markit, the momentum behind live deployments (in-use SDN) of SDN in the data center is expected to drive $11 billion in Ethernet switching revenue by calendar year 2022, with a CAGR of 27%. This is up from $2 billion in the first half of 2018, growing 12% from $1.9 billion in the second half of 2017.
Factors driving our forecast include:
- SDN network virtualization overlays (NVOs) become important technology for multi-cloud connectivity as enterprises deploy hybrid clouds and then expand them to include multiple CSPs.
- Telco deployment of SDN and NFV ramped in 2018 with deployment of mini data centers in central offices; this continues into 2019 and beyond.
- Use cases for SDN expand past automation to include disaster recovery, security, and business continuity.
- Deployment of switches with programmable silicon allowing protocol modifications continues to ramp into 2019.
Moving forward, we believe many Tier 2 cloud SPs, communication SPs, and enterprises will look to open networking for automating their multi-site data center networks in the near term and will expect to leverage the same open networking software stack as they move compute to the edge.
For an in-depth discovery of the world of SDN, including current and future innovations, view our recent webinar, Automating multi-site data center networks with SDN open networking, led by myself and sponsor Pluribus Networks. In this webinar, we dive into best practices, real-world use cases, common obstacles, and innovative solutions for what lies ahead.
Cliff Grossner, Ph.D., is executive research director and technology fellow in the Cloud and Data Center research service at IHS Markit
Posted 10 June 2019
- Network slicing before 5G: The benefits of slicing today
- 5G: “One network” capability is remarkable, but many challenges are still present
- Blockchain: Industries may not be prepared for the complex technology, but its impact will be huge
- 5G: The best is yet to come
- Artificial intelligence will be key to innovation and growth in system-on-chip solutions
- Network architecture changes are coming for 5G
- The state of the IIoT: Adoption hurdles and common pitfalls
- Artificial Intelligence ranks as the most impactful transformative technology in Digital Orbit
Artificial intelligence (AI) and machine learning (ML) play a vital role in the future of the Internet of Things (I… https://t.co/tBQYqOEtFw