Networking Challenges to the Distributed Enterprise

Even SDN won't provide everything that wide area data environments require. How can enterprises optimize their distributed infrastructures?

By Arthur Cole | Posted Dec 6, 2013
Page of   |  Back to Page 1
Print ArticleEmail Article
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn

Networking has always posed vexing challenges for the enterprise, even in the old days when entire data environments were confined on-site infrastructure. Now, with the front office looking to integrate branch offices, colocation providers, partners, suppliers and even customers under a unified, cloud-based infrastructure, the distributed enterprise is creating entirely new levels of network complexity.

Not to fear, though, since software defined networking (SDN) is coming to the rescue, right? Once you can dynamically configure end-to-end network infrastructure, won’t all your network issues vanish with just a few mouse clicks?

Not quite, according to NetSocket’s Dave Corley. Today’s multi-site enterprise has a host of requirements that cannot be addressed with current hardware-dependent SDN implementations. For one, they lack integrated DHCP, NAT and port-forwarding services to provide flexible, logical addressing for local subnets and locally hosted public servers. Additionally, without features like DSCP marking, priority queuing and inter-site tunneling, enterprises will have trouble maintaining multi-site QoS and security. Ultimately, this will lead to increased costs, greater instability and difficulty integrating legacy hardware into the new SDN infrastructure. A better approach, Corley says, is to virtualize all network functions, including routing, L3/L3 switching and tunneling, on commodity servers at each site. This way, you remove your dependency on specialized routing hardware, while at the same time preserving server resources for non-networking applications, ultimately providing a three- to five-fold cost benefit.

Any way you look at it, distributed architectures will place greater reliance on the edge, says Taneja Group’s Mike Matchett. The mistake most enterprises make once they go regional is that they try to manage the distributed network as a set of physical resources. They should instead shift the focus to data control. Through increased virtualization and intelligence on the edge, the enterprise will be better able to project centralized infrastructure, applications and other resources to remote sites, ultimately improving management and control and delivering better service to users.             

Part of this transformation will require the introduction of an Application Delivery Network, according to Research and Markets. With web-based collaboration, increased machine-to-machine (M2M) interaction and rising traffic from mobile applications, the ADN offers a means to push apps across distributed environments through a dedicated architecture not subject to the latency and bottlenecks that plague standard networking. Top ADN platforms also provide a wide range of classification and prioritization tools that provide real-time analysis of apps, content and even users to establish the proper levels of acceleration and security without compromising performance across the board. Many also offer network monitoring tools to avoid bottlenecks and minimize bandwidth consumption and resource utilization.

No matter how well architected, however, things can and do go wrong in the distributed enterprise. So enterprises must also establish a robust data recovery system capable of fulfilling the needs of a widely disparate infrastructure. Asigra, for example, has built its new Cloud Backup Connector Appliance around the Cisco ISR G2 router and the UCS E-Series blade to provide both local and remote recovery from external cloud resources. In this way, even globally distributed organizations can provide recovery services at LAN speeds, using an agentless approach that extends not just to storage, servers and desktops, but to laptops and mobile devices as well.

In a distributed enterprise, maintaining connectivity is not the central issue. Broadband networks have proven pretty reliable over the years. The real challenge lies in integrating disparate resources into a cohesive environment that maintains the control and security of centralized infrastructure while enabling LAN-like performance to even the most distant user.

All the tools are in place to achieve such a feat, but it’s up to each enterprise to figure out how to best deploy them over legacy distributed infrastructure.

Photo courtesy of Shutterstock.

Comment and Contribute
(Maximum characters: 1200). You have
characters left.
Get the Latest Scoop with Enterprise Networking Planet Newsletter