a choice of best-of-breed software and hardware components, and purchasing from any vendor that supports the common operational framework. In a disaggregated network an ODM manufacturer provides hardware built on merchant silicon while the software runs under a unified network operating system from a third-party NOS provider.
Hyperscale cloud providers have successfully pioneered this approach for compute and memory workloads. Powered by virtualization technologies, they have developed agile, dynamic and cost-effective infrastructure to run any application in their data centers.
Unique Opportunities for Service Providers, a Report by AvidThink:
a Disaggregation in Networking – The Second Wave
Learning Disaggregation From the Hyperscalers
Modern network operators have been struggling for years to increase agility of operations including quick service introduction, better visibility and control, less reliance on single vendors and decreased capital and operating costs. For that, some leading operators have tried to learn from the hyperscalers and have disaggregated some of their optical networks. They are also beginning to move to disaggregated architecture for low-scale/low-throughput services like edge-based vCPEs and firewalls. A few operators are even pushing disaggregation deeper into their network infrastructures for higher benefits.
At DriveNets, we believe disaggregated networks and routers will eventually prevail across all service providers infrastructure, for the following reasons:
- Technology is now mature to support any telco-grade network service, at any scale
- Disaggregation supports rapid introduction of technology innovations and new services
- Disaggregation’s proven working operational model
- Disaggregated routers deliver true savings compared to incumbent models
Disaggregation Technology Now Mature to Support Any Telco-grade Network Service – At Any Scale
Merchant silicon vendors such as Broadcom have successfully developed high-scale, carrier-grade chipsets with high density port interfaces (up to 400 GbE) and low power per bit (Jericho2 for example). This is critical for core networks where high reliability and high throughput are paramount. We believe that more vendors will follow with similar ASIC technologies.
In the control plane, routing functions and protocols are becoming available through different Network OS vendors, and for some of them like DriveNets, they have been redeveloped to benefit from the power of highly scalable, cloud-native architectures and work efficiently at scale.
Cloud-native architectures inherently provide support for high availability and resiliency through load balancing or fast container recovery. However, to reach the level of availability required by telco-grade routing, they also need to support fast data synchronization, state recovery and data integrity, at scale.
We are already starting to see global acceptance of the technology and architectures. AT&T, the largest operator in the world, is pushing for open, standardized white boxes by offering its DCC (Distributed Disaggregated Chassis) design to the OCP (Open Compute Project) community. The work done by TIP (Telecom Infra Project) for mobile access or ONF (Open Networking Foundation) is a strong sign of market acceptance as well.
Disaggregation Supports Rapid Introduction of Technology Innovations and New Services
This is crucial in today’s digital economy as network agility is important for adapting services quickly to the demand. Disaggregation also allows the faster development of new hardware- and software-capabilities independently from one another.
By leveraging container-based microservices, service providers can quickly add new services to the control plane without hardware or site limitation allowing them to expand service offerings quickly, effectively and at low cost. Again, the open and automated nature of disaggregated networks simplifies this expansion even further.
Disaggregation has a Proven Working Operational Model
While it may be unnerving not to have a single vendor providing all support needs, disaggregated products are maturing through testing, validation and certification activities both on the hardware and chipset levels. Likewise, the addition of new hardware models is validated through the interoperability verification of the software vendors.
Meanwhile, support and service operations are shifted from the vendor to the value-added reseller (VAR). This is now a one-stop-shop for commissioning, on-site integration, local installation, end-to-end testing and other functions in the installation phase. In the production phase either the software provider or VAR can handle all support and maintenance issues, and coordinates software upgrade and rollback operations for each vendor. In essence, this model becomes the same as today, but with the software vendor or VAR acting as the primary point of contact.
Disaggregated Routers Deliver True Savings Compared to Incumbent Models
For both CapEx and OpEx, disaggregated approach provides a budget-friendly infrastructure in terms of the initial deployment, growth and long-term TCO.
On the CapEx side, this new model offers lower hardware prices thanks to simpler hardware on a cost-plus model and healthy competition between different suppliers. Most carriers find they can implement significant consolidation of locations and services onto fewer white boxes with better hardware utilization metrics.
This consolidation also produces results on the OpEx side, as fewer boxes draw less energy for power and cooling, while overall network simplification and standardization reduces service assurance, network engineering, security, certification, upgrades and a wide range of other expenses.
This is why disaggregated routers are starting to enter leading Tier 1 operators. While this transition is still at its infancy, there are a few working projects showing very promising results.
Disaggregation Examples at Service Providers
AT&T, for example, recently announced that they are on track to have 75% of their core network functions fully virtualized and software-controlled by end of 2020. It is not only happening for their 5G cell tower routers announced in 2018 but also for most of their network functions. The disaggregation model is the core enabler for it.
British Telecom, meanwhile, is turning to disaggregation as a means to support next-generation container and cloud-native deployments for both mobile and broadband applications. Ultimately, the company is looking to converge broadband, Wi-Fi, cellular and other technologies using both homegrown and third-party solutions to create a truly unique network ecosystem.
From DriveNets’ own experience, we are seeing more and more interest in disaggregated routing solutions from operators around the globe who are looking to drive enormous network scale while lowering their TCO.
DriveNets is no doubt, at the forefront of this revolution, building large but simpler networks for operators so they can deliver better services to their customers.