The future of ATM and frame relay in an IP world

Remember the debates of IP vs ATM? Those passionate arguments for ATM in the LAN and the WAN eventually withered from the IP blitz. Given the direction that carriers are moving, is ATM dead? Far from it.

Sure, investments have been declining after many years of growth, but the market still has legs. As part of the overall market downturn, the multiservice (a.k.a. frame relay and ATM) edge switch market slipped to $1.2 billion worldwide in 2003 from a high of $3.8 billion in 2000. It is still a sizable market, and Infonetics Research projects that it will stabilize and begin to grow moderately again in 2004 as carriers look for cost reductions and greater port density in multiservice switches, invest in growth areas requiring ATM technology, and prepare for data network convergence over MPLS backbones. The five main driving applications for multiservice switch sales in the next three to five years are listed in order of importance below. All applications will involve Ethernet or IP in the end, but in the near-term, multiservice switches have a role.

DSL aggregation: This is the chief driver for continued investment, and as incumbent carriers ramp up their DSL subscriber base and prepare for fiber deployments, carriers will need to put more money into the broadband aggregation network, which is primarily based on ATM switches. Investment in multiservice switches for DSL aggregation is driven mostly by DSL subscriber growth, the increase in connection speeds, and the broadening of the services delivered over DSL. This will definitely drive multiservice switch investment for the next five years and is the overall prime driving application. However, this is not an unlimited growth opportunity: as emerging IP-centric DSL is deployed in greenfield environments and Ethernet trunk links are added to ATM DSLAMs, this will diminish the opportunity for new multiservice switch aggregation networks as carriers begin to use router and Ethernet switch aggregation networks.

FTTx aggregation: ILECs are not stopping with DSL as part of their broadband strategy, as they have a two-pronged FTTx rollout strategy that involves fiber to the neighborhood (FTTN, a.k.a. fiber to the node) and fiber to the premises (FTTP). Initial FTTP deployments are based on APON and BPON; both are ATM-based FTTP technologies that require an ATM aggregation network for circuit aggregation. Multiservice switches will see modest lift over the next four to five years from FTTP deployments, but, like DSL, carriers will ultimately cap those investments as they evolve to Ethernet-based aggregation networks to support the GPON and IP DSLAMs (at the end of FTTN loops) that will make up most new deployments.

Frame relay and ATM data services: The frame relay and ATM service markets are mammoth and generally projected to grow, and entrenched service providers buy new equipment in line with continued traffic growth. There are two main trends with this market in the US: RBOCs are investing in Layer 2 data networks as they support network, customer, and revenue growth, which is putting competitive pressure on IXCs with their declining customer base and slowing traffic growth leads to dwindling multiservice switch investment. Also, IP VPN services are growing in popularity, and carriers are looking to migrate customers to these new services. As migration is not a flash cut scenario, carriers are creating service offerings involving both frame relay and Layer 3 IP VPNs and frame relay will exist as an access option for VPN service for years to come. But, in the long run, Layer 3 IP services are a threat to frame relay and ATM services and will detract from investments in multiservice switches.

Mobile voice and data backhaul: Multiservice switches are used for access aggregation and as backbone networks for mobile voice (2/2.5/3G) infrastructure. There is more opportunity in these greenfield situations, as these networks are in growth phases in terms of capacity and footprint. Multiservice switches are used for aggregation of mobile voice and data traffic in 2G and 2.5G networks and for the initial designs and requirements for 3G networks. Although the emerging 3G mobile access networks are based on ATM, like the broadband aggregation networks, the long-term vision is to use IP in mobile networks, putting a damper on long-term growth for multiservice switches for this application.

MPLS migration and data network convergence: Over the past 10 years, service providers have made significant investments in large frame relay and ATM networks, and they still need to offer these services due to customer demand. However, they want to prepare themselves for IP service growth and interwork their current slow growing, yet profitable legacy data networks with the same networks carrying their high growth, low margin IP services. Carriers are preparing for a massive consolidation of their networks over one unifying MPLS backbone, interworking with many legacy and emerging access networks. For multiservice switches to thrive in this environment, they must 1) enable new services over the existing multiservice infrastructure, and 2) enable the staged migration of existing services to MPLS infrastructures. The role of multiservice switches in this role is by no means assured, as many carriers seem inclined to deploy new IP centric equipment (i.e., routers) for interworking and as a gateway to the MPLS core. It is too early to decide which approaches—router or multiservice switch—will win in the market, and in the next two to three years, multiple approaches will likely be used, driving switch sales.

New opportunities for multiservice switch sales in developed markets are few, and most growth is through card additions to existing footprints, yet there are developments that breathe new life in multiservice switches. Furthermore, in high growth regions like Eastern Europe and Asia Pacific, there are more greenfield opportunities for all five main driving applications described above. In the end, all these driving applications have their length shortened by Ethernet or IP, but in the next three to five years, they will sustain the multiservice switch market, giving ATM and frame relay a future in a world oriented toward IP.

Kevin Mitchell is the directing analyst with Infonetics Research and can be reached at kevin@infonetics.com.