Speak Like a Data Center Geek: Networks


It’s literally true that we can’t get enough of networks at Equinix. We were founded as a neutral place where network service providers could come together and exchange data traffic, and today we host more than 1,000 networks, with room for more. That all makes networks a solid choice to headline the 10th installment of our “How to Speak Like a Data Center Geek” series, which aims to bring clarity to the sometimes opaque terms we data center types toss around.

Network Neutrality

This concept is at the heart of the Equinix business model. We don’t favor any single network service provider (NSP) and we don’t own any networks. We are all about providing a place for colocation, interconnection and cost-effective peering for any NSP, which include Tier-1 carriers and Internet service providers (ISPs). Our neutrality has drawn a wide range of NSPs who want to be a part of expanding network ecosystems, and they, in turn, have drawn other providers (such as cloud service providers) and businesses who want access to them and each other. We’ve been called the Switzerland of data centers, and now you know that’s got nothing to do with the chocolate in our vending machines.

Wide Area Network (WAN)

A “wide area network” is better known by the acronym WAN, not to be confused with a wan winter light, or any other poetic wans. An organization’s WAN is a long-haul network that transmits data among employees, customers and partners in different locales. A key is that a WAN consists of the leased telecommunications lines (i.e., T1/T3 links) that NSPs provide, as opposed to a LAN (local area network) or MAN (metropolitan network), which organizations use to connect people in much smaller areas with mediums that have more limited range than leased lines, such as Ethernet.

Software Defined Networking (SDN)

SDN makes a network more flexible and efficient by freeing the centralized control of the network from each physical network device. In SDN, the “control plane,” or the brain inside of network hardware (switch, router, firewall, etc.) that decides where traffic is sent, is decoupled from the “forwarding plane,” or the muscle that forwards the traffic. The control plane is replaced with a software application called a “controller.” The programmable controller provides end-to-end visibility of the entire network, and centrally and intelligently manages the flow of the data to and from each device, such as by finding the fastest route to wherever. In a SDN, a network administrator can shape traffic (prioritizing, de-prioritizing or blocking specific types of packets) from a centralized control console without having to physically touch each individual network device, which is both efficient and cost-effective.

Network Functions Virtualization (NFV)

The Network Function Virtualization concept is, as the term suggests, about replacing the various functions (distributed naming services, caching, etc.) typically performed by higher-end network hardware with software running on virtualized commodity servers. NFV allows network operators to save both capital expenses (CAPEX) and operating expenses (OPEX) by adding specific network functions to existing servers via software. It also enables them to avoid committing to the space, power and cooling needs required by dedicated networking equipment. Besides being cheaper, NFV enables network providers to use only the compute resources needed for a specific task and, if more is required, they can virtually add it on the spot.

Load Balancer

A load balancer is the traffic cop of the network world. It distributes network and application traffic across a number of servers or, in some cases, networks, so that no one application server or network becomes a potential single point of failure or bottleneck. By distributing the traffic, a load balancer also decreases the burden on these servers and networks, improving application reliability, performance and network service assurance.

Maybe it’s obvious after a post like this, but we data center geeks have a thing for interconnection, since it’s essential for the enterprise to compete. Download Equinix’s IOA Playbook, which describes an interconnection-first architecture that securely connects people, locations, clouds and data.

And check out every post in the “Speak Like a Data Center Geek” series. (Please note: We welcome binge readers):

Part 1: Introduction

Part 2: Power I

Part 3: Connections I

Part 4: Cloud

Part 5: Buildings

Part 6: The stuff we sell

Part 7: Security and reliability

Part 8: Connections II

Part 9: Sustainability

Part 10: Networks (See post above)

Part 11: Power II

Part 12: Internet of Things

Part 13: Big Data

Part 14: Virtualization

Part 15: Virtual Reality

Part 16: Software Containers

Part 17: Artificial Intelligence

Print Friendly

Related Content