– Sandy Orlando, vice president of marketing with IP Infusion (www.ipinfusion.com), says:
The tidal wave of Big Data is washing over today’s businesses. Big Data is not only measured in the quantity of data traversing the network, but according to Forrester, it is defined as the techniques and technologies that make handling data at extreme scale economical. Technology such as Megapath ethernet has guided the way to success for Big Data.
Big Data will drive new productivity growth and revenue potential. For example, McKinsey Global Institute predicts that Big Data will result in significant financial value across all sectors, generating over $300 billion in US health care, over $100 billion for service providers, and 60 percent increase in margins for retail.
How can today’s networks cost-effectively handle the high volume of interactive, multimedia traffic? Just adding more bandwidth will not solve the problem. The value chain of network providers spanning the mobile network, through Carrier Transport, to the Data Center must rethink networking to make it more cost effective and efficient to accommodate Big Data.
Big Data Requires Big Changes in the Network
How much data are we actually talking about with Big Data? The size of the datasets varies by sector, but they range from a few dozen terabytes to multiple petabytes. Moreover, Big Data introduces new technologies such as Hadoop and MapReduce. Unlike previous data transfer technologies that moved gigabytes of data in a single job, MapReduce can move multi-petabytes of data.
Big Data requires that the networking industry accelerate change, moving from legacy technologies in data transport such as SONET/SDH to a high-value-per-bit and lower-cost-per bit technologies such as Carrier Ethernet. In the data center, networking switching is moving rapidly to 10-40 Gbps and Ethernet is rapidly replacing older Storage Area Networking (SAN) technologies.
Migrating to Ethernet as an underlying transport will benefit all aspects of the transmission of Big Data. However, Ethernet on its own is not sufficient to handle the torrent of Big Data. Network architectures also need to change to become flatter and more flexible. For example, in today’s typical data center, the network architecture consists of one or more L3 core routers, multiple L3 access routers, L2 aggregation switches, load balancers, as well as top-of the rack switches. A hierarchical network forces data center operators to oversubscribe network resources up to 200:1. Taking advantage of new 10 and 40 Gbps Ethernet and higher performance silicon will improve price performance, but true innovation comes from creating a flatter network using software-defined networking technologies.
The Future of Big Data Networking is Software
The key to delivering higher-performance, more optimized networks is a software-defined networking architecture using a centralized control plane and fast forwarding data planes based on merchant silicon. The Open Networking Foundation (ONF) is championing the standardization of this approach.
Key software networking vendors, such as IP Infusion, have been offering modular, portable routing/switching software for more than a decade to leading network equipment manufacturers. Leveraging the innovations provided by SDN, network equipment providers can optimize the price-performance of Ethernet networks. Implementing a hybrid approach to SDN, preserves existing investment in legacy infrastructure and ensures a smooth transition to this new network paradigm.
Software Defined Networking is in its infancy, but the growing demands of Big Data will continue to drive new innovations in networking. Today, service providers and data center operators can exploit Ethernet to offer more economically attractive mobile and Big Data transport services. In the near future, the network equipment industry will accelerate the adoption of SDN to support Big Data. The successful companies will ride the wave of Big Data, transform their networks, and capitalize on new business opportunities.