Juniper Networks announced today, March 3, that the networking company was releasing a new suite of products and services that delivers what its calls the "first Converged Supercore" for service and content providers: Juniper's Converged Supercore This is an integrated end-to-end packet transport solution optimized to deliver the switching and transport technology.
There's no question that there's a need for a next generation of core networking technologies for Internet Service Providers (ISP)s and content delivery networks (CDN)s. We have the basic technology now in IEEE 802.3ba, which sets down the technical guidelines for 40 Gigabit Ethernet (GigE) and 100GigE Ethernet and we also have the customer demand for that kind of speed.
As Juniper explained in a statement, "With nearly one billion mobile smart devices and an increasing shift toward cloud-based and video applications generating exponentially growing and unpredictable traffic, service providers have to evolve the core network to flexibly accommodate new levels of peak demand while improving the economics of the core network infrastructure." Juniper is right.
There are already times when Netflix online video alone takes up most of the available Internet broadband. ABI Research states that in 2010 Wi-Fi penetration in US commercial establishments employing more than five people grew by 45% by the end of 2010. And, ABI also reports that Mobile operators in dealing with the mobile data traffic boom, as smartphones and tablets redefine the end-user mobile experience. As Aditya Kaul, Mobile Networks practice director at ABI Research said, "Operators are investing significantly in their core network architecture, not just to speed up the throughput on their networks but also to make those networks more efficient and 'aware' of the traffic passing through them."
So it's a perfect time for Juniper to try to snatch the brass ring of high-end Internet switches from Cisco. The real question is can they do it?
In a statement, Kempei Fukuda, Senior Director, Global Network, NTT Communications, a leading Japanese telecomm and datacomm company indicates that NTT thinks Juniper has a shot. "The new flexible approach to network architecture that Juniper is advancing accommodates the increasingly unpredictable traffic patterns, driven by new and emerging cloud computing, video and mobile applications. By creating a hybrid approach to the packet and optical networks, Juniper has introduced a cost-effective converged solution that dramatically reduces complexity while building in the scalability we need to quickly adjust to unforeseen traffic demands," said Fukuda.
Here's how Juniper claims it can handle those data spikes. Instead of using rigid circuit switching model, which can break when confronted by unpredictable, burst-prone traffic patterns, Juniper's new architecture blends optical transport with Multiprotocol Label Switching (MPLS) in a Converged Supercore, making the best use of both technologies. To do this it uses a single operating system, Junos, to power both the optical layer and the packet layer.
These new high-end switches, the Juniper Networks PTX Series Packet Transport Switchseries, also uses a new chipset: the Junos Express chipset. Juniper claims that these chips will the make its switchs' slots the "industry's fastest, most energy efficient and most scalable at up to two Terabits per slot."
All this, and Juniper also claims that their Converged Supercore will also deliver "network capex [capital expenditures] cost savings of 40 to 65 percent compared to traditional architectures and a 35 percent savings versus a pure IP routing solution."
Those some claims, but I'm willing to be convinced. As ISPs and CDNs both know, they already need this kind of performance at affordable price points, and that's before the coming surge in Internet traffic demand arrives from yet more mobile and video-happy users. Like it or not, Internet providers are going to need to invest in new, higher-end core equipment and Juniper's new hardware should be on their short-list for consideration.