Business rationale for IP/optical integration
IP/optical integration is a key strategy to address both short term and strategic service provider business challenges.
In the near term, IP/optical integration removes operational and technological barriers that inflate overhead cost and impede routing and transport convergence. Service providers will more economically scale network capacity, efficiently conduct cross-domain network operations, and effortlessly monetize network assets.
For the longer term, IP/optical integration provides the agility and programmability needed for innovative control paradigms based on software defined networking (SDN). SDN further improves network utilization with on-line traffic engineering and bulk optimization to dynamically adjust capacity to fluctuating demand. SDN also enables emerging cloud service offerings such as public and private datacenter interconnect.
IP/optical integration addresses the disparities and disconnects at the management, control and data plane that make conventional routing and transport networks so costly and cumbersome to operate (Figure 1, left).
- Legacy transport services are statically provisioned.
- Manual intervention and elaborate workflows are typically needed to orchestrate cross-layer operations.
- Service provisioning is complex, time consuming and error prone.
- The network infrastructure is far too rigid, static and disconnected to adequately respond to fluctuating demands, or to facilitate dynamic service control.
Evolving to a dynamic, agile and programmable network
State-of-the-art routing and optical transport technologies foster a far more integrated deployment model. Service providers can leverage these technologies to transition from operating IP routing and optical transport in separate silos, to an integrated multi-layer deployment model with significant cost and performance synergies.
IP/Optical control plane integration is the most strategic aspect. It can be conducted in parallel with management and data plane integration as each plane addresses different issues and offers complementary integration benefits. The following 3 steps establish a dynamic, programmable and unified multi-layer control plane (Figure 1, right):
- Equip the optical transport network with a GMPLS control plane. Optical Transport Network (OTN) and reconfigurable optical add-drop multiplexors (ROADMs) technology offer flexible and cost-efficient grooming and transport for IP/Ethernet and SDH/SONET payloads. GMPLS protection and dynamic restoration features leverage this flexibility to more efficiently protect services and improve network utilization.
- Extend dynamic transport control capabilities to the routing layer through a GMPLS User-Network Interface. GMPLS UNI integration improves operational efficiencies by closing the loop between the routing and transport control planes without requiring manual intervention. It establishes a unified multi-layer control plane to coordinate traffic forwarding, protection and restoration in the most economical way.
- Expose an SDN abstraction layer with programmatic SDN application interfaces. to help monetize the multi-layer network as virtualized, cloud consumable services. SDN control elements such as the path computation element (PCE) help optimize capacity usage by enabling dynamic traffic engineering and bandwidth calendaring.
Each integration step delivers immediate cost and efficiency benefits and contributes to an open and programmable multi-layer control plane to improve service velocity and facilitate dynamic SDN control. The next three sections illustrate the benefits of this.
Dynamic protection and restoration with GMPLS
Service providers are always looking for ways to run their networks hotter and maximize returns on network investments. Conventional transport networks often reserve as much as 50% (referred to as 1+1 protection) of provisioned capacity to recover from failures. This approach is very costly and even ineffective for protecting IP traffic:
- Besides keeping up to 50% of transport capacity in reserve, backup protection paths are pre-provisioned, unable to take the actual failure location into account for re-routing options, and unable to recover from multiple failures.
- Most IP traffic can tolerate brief outages and limited packet loss. This traffic can leverage the alternate network paths instead of dedicated 1+1 protected paths.
- IP routing topologies typically have high degrees of connectivity with many alternate paths. However, 1+1 protection only supports the use of a single backup path.
State-of-the-art OTN and ROADM transport technologies can use an intelligent control plane that provides the right amount of protection for a given service class, leading to more cost-effective utilization of networks resources. These protection capabilities leverage the generalized multiprotocol label switching (GMPLS - RFC 3945) architecture. GMPLS adopts key concepts from the MPLS control plane used in IP routing with functional enhancements to support multi-layer transport networks.
GMPLS enables the transport network to dynamically route or reroute traffic around failures or on to optimal paths based on network utilization and/or service SLA constraints. The GMPLS user-network interface (UNI) interface lets routers dynamically signal transport paths with support of various service protection options (Figure 2).
Dynamic restoration capabilities enable efficient sharing of backup resources by moving from a 1+1 to a shared alternate path protection model that is also able to recover from multiple failures. As a result a considerable amount of reserved protection resources are freed up, with the remaining being used for revenue generating traffic with strict SLA criteria, allowing for more capacity for less demanding services while also offering these services high availability via alternate path based protection.
With the right architecture, GMPLS-based transport layer recovery mechanisms can be applied in combination with recovery mechanisms in the IP/MPLS layer to implement differentiated availability SLAs for different classes of service. Differentiated service availability requirements can be mapped on an appropriate multi-layer traffic protection and restoration strategy in order to balance availability, redundancy and resource utilization for the best returns on network investments.
An internal Alcatel-Lucent study shows that GMPLS protection and dynamic restoration capabilities can save 30 to 50% on router ports and optical transponders, compared to using conventional 1+1 optical protection and MPLS fast reroute. These savings are achieved with comparable or better network availability and resilience.
Simplify and streamline cross-domain operation with GMPLS UNI
Most service providers take a segregated approach to routing and transport operation, which results in a fragmented management view. This makes it:
- Cumbersome to manage connectivity between routing and transport systems
- Time consuming to correlate and isolate faults that impact multiple domains
- Labor-intensive and error-prone to orchestrate resource operations across domains
- Costly to automate operations due to the need to integrate and coordinate different protocols, information bases and processes across routing and transport devices
The GMPLS User to Networking Interface (RFC 4208) enables crossing the administrative boundary between routing and transport, allowing the IP routing layer to communicate its resource requirements to the underlying transport network without the need for operator intervention (Figure 3). The reverse interaction is also enabled. The transport layer can directly inform the IP routing layer about relevant events to more effectively use network resources.
Closing the loop between routing and optical transport enables a consistent use of network resources, which leads to improved network efficiency because it removes the need to touch each layer separately. It also helps to consolidate, coordinate and automate management activities across routing and transport layers with fewer touch points and less manual intervention.
Capitalizing on dynamic SDN resource control
Conventional bandwidth provisioning practices only follow the long term trend. Usage statistics over the past weeks or months are collected, aggregated and fed into off-line traffic engineering tools that estimate what capacity adjustments are required to fulfill demand for the coming weeks and months.
With statically provisioned capacity based on forecasts, service providers must err on the side of caution by over-provisioning capacity. Unexpected demand surges or traffic spikes can still lead to traffic being clipped at occasions, without opportunities to leverage or share unused capacity during off-peak hours for other tasks (Figure 4, left).
Introducing more agile capabilities that can make periodical adjustments to allocated capacity can tap into a number of new revenue generating opportunities that arise from bandwidth calendaring and off-peak/on-peak hour services. For example, database auditing or backup services could benefit from unused capacity during off-peak hours. The ability to make periodical adjustments to network capacity can improve network utilization, for example by utilizing the complementary peak hours of residential and commercial traffic to lower the total amount of peak capacity needed (Figure 2, right).
The ability to make even more agile and dynamic capacity changes would enable the introduction of more efficient bandwidth-on-demand and bandwidth bursting services through self-service portals, two service capabilities that are highly desired by business users. Applying SDN for dynamic load-balancing, on-line traffic engineering and bulk optimization can mitigate network hotspots and better align capacity demand and availability. A converged IP/optical network with dynamic SDN resource management can adapt to unexpected demand fluctuations with minimal capacity requirements.
A strategy for IP/optical integration White PaperThe optical networking TechZine eBookAgile Optical Networking webpageIntegrated Packet Transport webpageIP-core-routing solution webpageMetro solution webpage
-  Notable interfaces are OpenFlow, BGP-LS and the PCE Communica¬tion Protocol or PCEP (RFC 5440).
To contact the author or request additional information, please send an email to firstname.lastname@example.org.