What is a Multilayer Path Computation Element and How Does it Speed Cloud Migration?

How Centralized Multilayer Path Computation Optimizes Circuit Design

Transport networks are typically composed of thousands of interconnected devices from multiple vendors. This includes a mix of physical and virtual Network Elements (NEs), domain controllers, and service orchestrators. With the continued evolution and migration of service provider OSS/BSS systems and network infrastructure into the cloud, a centralized Multi-Layer Path Computation Element (ML-PCE) plays a critical role in managing circuit design for transport networks, and in establishing circuits across network devices. A Centralized ML-PCE is a key part of the puzzle in enabling enterprises to operate efficiently and effectively while consumers enjoy the modern age connectivity perks of Amazon Prime, Hulu, Netflix, YouTube, social media and more. 

Centralized ML-PCE Optimizes Circuit Design for the Cloud-Native Environment

There are multiple approaches for path computation, but Fujitsu’s centralized ML-PCE is highly adapted to the cloud-native environment. As operators move towards cloud adoption and networks grow in scale and complexity, it becomes critical to constantly optimize traffic routes and anticipate future connectivity demands. Working alongside network management systems and SDN controllers, the centralized ML-PCE determines the best route for traffic and signal paths, and it optimizes across WDM, OTN and packet layers as necessary.

Fujitsu’s centralized ML-PCE uses enhanced and customized state-of-the-art algorithms for conveying data between a source and a destination, making it possible for data to easily and rapidly flow through the network.This hides the proprietary details from the user and enables seamless and quick circuit design and activation across the network.

Three Key Differentiators for Fujitsu Centralized ML-PCE

Three key differentiators set centralized ML-PCE apart from other approaches:

• Graph database and massive processing power speeds up circuit design and activation
• Fujitsu ML-PCE is embedded in the Virtuora Network Controller (NC) platform, making integration easier with standard APIs
• The Fujitsu ML-PCE Architecture Handles Complex Constraints in a Modular, Multilayer, Multivendor Environment

Graph Database and Massive Processing Power Speed Circuit Design and Activation

Quick path computation is critical to circuit design in cloud-native transport networks. Coupled with Fujitsu’s Virtuora Network Controller, centralized ML-PCE uses an advanced graph database with incredible processing power to identify routes within milliseconds (depending on network size and complexity). Graph databases help quickly compute the path and navigate through the network faster than any relational database. As business processes migrate into the cloud, a modular approach using a centralized graph database allows network operators to seamlessly enhance computing needs on-demand as their network grows, supporting a PAYG model.

In addition to regular service creation, controller-based path computation dynamically calculates restoration paths. This automated process requires no user intervention; consequently, downtime is dramatically reduced. With a centralized ML-PCE approach, operations can seamlessly centralize the planning of restoration routes and prevent collisions and contention throughout the network. This is one of the factors in faster, uninterrupted transmission, and it ultimately speeds service delivery. 

Virtuora NC’s Standard APIs Support ML-PCE with Easier, Faster Integration

The second differentiator is that the Fujitsu centralized ML-PCE is embedded in the Virtuora NC Software-Defined Networking (SDN) platform. The benefit is abstraction of network complexity and the use of standardized, open APIs for northbound and southbound integration. This makes software and hardware integration easier across multiple vendor solutions. For example, Virtuora YANG models automatically make Southbound Interfaces (SBI) and Northbound Interfaces (NBI) available, simplifying both NE configuration retrieval and provisioning workflows.

The Fujitsu solution is open source, built with OpenDaylight (ODL), and it supports a multivendor model for connectivity. Centralized ML-PCE works with legacy equipment as well as current technology, and allows for a mix of equipment and vendors while still supporting both service creation and restoration. This is not possible with a distributed control plane. Centralized ML-PCE also allows users to rapidly take advantage of advances in computing power without needing to upgrade all of the devices in the network. The Fujitsu centralized ML-PCE also supports any multivendor OLS model, enabling seamless support and upgrades. As a result, Operations can control network connectivity resources on demand and automate configurations and service deployments.

Centralized ML-PCE Handles Service Constraints with Minimal Performance Impact

The third differentiator of the Fujitsu centralized ML-PCE is its modular multilayer, multivendor architecture. This architecture can handle numerous concurrent route constraints with minimal impact on performance. This makes for a feature-rich ML-PCE solution that is easier to troubleshoot. It also provides numerous complex service diversity constraints, such as the ability to exclude or include links, nodes, node+SRLG, SRLG, site, and site+SRLG. Specifically, a centralized PCE allows custom routing constraints that are less feasible with other technologies such as distributed PCE. Centralized ML-PCE coupled with Virtuora NC helps operators improve connectivity management for heterogeneous networks. It also allows enterprises to manage and interoperate multivendor devices with one solution, preventing vendor lock-in, while evolving their infrastructure into the cloud.