JNT Visual - Fotolia

Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Overlay networks could revolutionize ISP backbone

Overlay networks have already taken the data center by storm, but they also offer advantages to the ISP backbone network, including more agility and reduced Opex.

Editor's note: In part one of our three-part series on deploying network overlays in Internet service providers' backbones, networking expert Jeff Loughridge explores the history of overlays in the backbone network. Parts two and three look at the specific ways an overlay network with an MPLS fabric beneath it could benefit an ISP's backbone, without a total rip-and-replace of the existing infrastructure.

Network overlays are all the rage in the data center, where the underlying network can act as the foundation for the more complex, virtualized network services on top. But the value of overlay networks doesn't stop there. Moving beyond the data center, Internet service providers (ISPs) are looking at network overlays to address different requirements and challenges in their backbone networks.

Existing networks are static and inflexible; provisioning services and implementing other changes currently takes weeks or even months. Overlay networks, however, could result in dramatically increased agility and significantly reduced Opex.

For an ISP, employing overlay networks based on familiar tunneling technologies -- as opposed to virtual LANs -- offers some intriguing possibilities. A look at the history and emerging uses of network overlays shows what may be possible.

The idea of virtualizing networks isn't new, particularly at the OSI reference model's layer 2 as applied to Ethernet. Long before the development of "fancier" layer 2 segmentation -- which includes Provider Backbone Bridges (also MAC-in-MAC) and Q-in-Q (also provider bridging) -- virtual LAN (VLAN) tagging was used to limit the size of broadcast domains. VLANs were ubiquitous in enterprise environments and remain so. But the complexity of VLAN-heavy environments hinders the scalability of these networks, as evidenced by the avoidance of VLANs in many hyperscale Web data centers.

Tier 1 ISPs have traditionally steered clear of VLANs, with minor exceptions like Network-to-Network Interfaces and their own data centers. The links that compose the backbone are almost always layer 3 point-to-point links that provide natural layer 2 domain boundaries.

Still, while large providers have generally avoided VLANs, notable exceptions exist. Many mobile broadband networks -- and their previous 2G incarnations -- were built by enterprise architects with limited exposure to best practices employed by the Tier 1 service providers. The heritage of these networks is apparent, as VLANs proliferate both in the mobile core and access networks.

MPLS/VPN network overlays change the game

The earliest incarnation of layer 3 overlays in ISP networks appeared in the form of tunnel-based virtual private networks (VPNs) in the mid to late 1990s. ISPs in the managed network provider space often used GRE or IPsec tunneling to enable the virtualized customer networks. The need for this VPN service was clear, but the implementation had numerous disadvantages. The routing complexity increased for both provider and customer, and turning up new customer sites involved manual and error-prone procedures.

The early 2000s ushered in a new era of layer 3 VPNs in the form of MPLS VPNs. Multiprotocol label switching (MPLS), developed in the late 1990s, became popular with many tier 1 ISPs. The standardization of the MPLS VPN architecture in RFC 2547bis gave providers that had already adopted MPLS a scalable architecture for offering layer 3 VPN services. MPLS VPNs offloaded routing complexity from the customer to the provider, giving the customer the appearance of having sites connected to a private IP backbone. They had none of the restrictions on the use of IP address space that had accompanied certain flavors of the original managed VPNs. The customer could use routing protocols like BGP and OSPF to communicate with the provider's routers, a capability that eased the introduction of the MPLS VPN backbone as the virtual enterprise backbone.

MPLS VPNs proved an enormous success. The time of their emergence was extremely opportune. Enterprises were in the midst of moving away from frame relay, ATM and private lines in favor of IP. MPLS VPNs naturally filled the gap for enterprises that wanted layer 3 connectivity between sites but were uncomfortable with using the Internet.

If the enterprise can lay claim to the first widespread success in network overlays in the form of VLANs, then ISPs can point to the MPLS VPN as a hugely successful network overlay that has its genesis solely in the service provider community. MPLS VPN remains entrenched in provider core networks and has extended into metro and access networks. Now the promise of network overlays in the data center has ISP executives and architects considering how the benefits of overlays -- primarily in flexibility and orchestration -- can be realized in the backbone beyond MPLS VPN service.

Next: In part two of our series on network overlays, we examine specific ways they could benefit ISPs, and how providers might implement them without ripping and replacing the backbone network.

Next Steps

Why overlay networks have made a splash in data centers

Learn more about service providers' adoption of MPLS VPNs

How an overlay network and a software-defined network differ

This was last published in April 2015

Dig Deeper on Data Center Networking

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

Do you think Internet service providers' experience with MPLS VPNs makes them more likely to implement network overlays in the backbone?
Cancel

-ADS BY GOOGLE

SearchUnifiedCommunications

SearchMobileComputing

SearchDataCenter

SearchITChannel

Close