When O'Neal Steel began rolling out its new ERP system, 28 remote sites needed fast, reliable access to the Oracle EnterpriseOne software running on a data center host. As a result, bandwidth demands increased fourfold or more, which caused serious response-time problems on the company's frame relay network, reports Michael Gooldrup, vice president of IS at the privately held midmarket metals processor.
So a move to MPLS service became pretty much a no-brainer.
"Frame relay offers no way to prioritize bandwidth; all applications compete equally," says Gooldrup, who finished rolling out AT&T's Multiprotocol Label Switching (MPLS) offering at his Birmingham, Ala.-based company in February. MPLS enabled O'Neal Steel to define a class of service that guaranteed adequate response time for the enterprise resource planning (ERP) application. Frame relay doesn't offer that capability.
O'Neal Steel's situation is far from atypical. For corporate America as a whole, and midmarket companies in particular, the move to MPLS is more a matter of when than if, according to Lisa Pierce, a vice president at Forrester Research. "MPLS is more than an upgrade path for frame relay or ATM [Asynchronous Transfer Mode]: It's the basis by which we'll be doing all kinds of transactions," from Voice over Internet Protocol (VoIP) to disaster recovery to service-oriented architecture, she says.
In traditional IP backbones, each router maps out the next segment of a packet's journey based on current traffic conditions and least-cost routing algorithms. This minimizes congestion and latency for the network as a whole but doesn't ensure that a particular packet stream gets from point A to point B by a specific time.
This was first published in August 2006