When I read press releases like this I laugh: "Despite the mounting frustration with MPLS connectivity costs and inflexibility, many customers still do not believe the Internet can provide the security and performance needed for today's business-critical traffic." MPLS is inflexible? It isn't inflexible; it is just complex.
Over the last seven years, I have watched many network admins try to wrestle with MPLS. Virtual pathways on a private network have been around since the days of frame relay. PVCs and SVCs were just traffic pathways in the frame relay cloud. ATM was more complicated to explain. "Frame relay won in the WAN over ATM, which proved too expensive despite its good points, such as five levels of QoS." The same will be said of Ethernet over MPLS - too expensive and complex despite its good points, such as security and QoS.
I think folks forget when you had to buy CIR (committed information rate) on WAN ports. VLANs are just dedicated bandwidth pathways for selected traffic. You want real-time traffic to have priority over email and web traffic. VLANs create that. The VLAN is not dynamic, so that might be why admins are leaning towards ethernet in the WAN.
Plus network admins are comfortable with ethernet. It has been the default for LANs, since before many of them were born. Not many remember ArcNet and Token Ring -- and the pain of running those LAN protocols.
Personally, IP-VPN is just adding more traffic to the Internet which is congested enough as it is. Over the top video buffers and OTT VoIP is jittery and cuts out like a cellphone call with bad reception.
WAN Optimization companies want to replace MPLS with broadband. That should work out well. The network admins will be happy because it will simplify the WAN. It will add security issues in a time when everyone is being hacked including the federal government and even security firms like Kapersky and LastPass.
The biggest complaint I get about MPLS is the expense. It does cost more than the Internet, but then it is more secure and dedicated bandwidth end to end. The configuration of the network is also complex, but you can outsource that.
"According to analyst firm Gartner, "Most network managers are feeling the pain of an increasingly complicated WAN with costly and time-consuming branch office solutions." This is true, especially for satellite offices of less than 10 people. However, if retail chains can utilize MPLS over DSL for POS, credit card processing, inventory and VoIP, why can't other industries?
The new thing is SD-WAN - the software defined WAN. This is basically on-demand bandwidth. Dial it up, dial it back, pay for usage. The CEO of Cogent used to say, "Just throw more bandwidth at it" as the answer to any problem on the network. This is a way to throw more bandwidth at an immediate problem.
SD-WAN will likely win, because it will put control in the network admin's hands. It is Ethernet. TW Telecom already had this option on some routes (where they owned the fiber). The added cost wasn't that heavy.
Frame Relay died. It was fairly easy. (Certainly easier to explain than ATM.) IP-VPN waned. Networks were segmented; now they are converged - with not just voice and data, but voice, data, Internet and video. IP-VPNs are coming back because Metro Ethernet has made the WAN pipes fairly inexpensive (moreso than DS3's!). WAN optimization and software defined networks will give the IP-VPN a revitalization.