What CIOs should know about the cloud-era network ‘tsunami’

Data centers and networks transformed to take advantage of the benefits of virtualization and cloud computing have become the new normal.

Servers and storage are being virtualized. On the leading edge, network function virtualization is well under way as software-based firewalls, load balancers and switches run on any cloud-enabled hardware.

Virtualization will enable rapid deployment of cloud services across the enterprise network, resulting in the network and data center becoming more intrinsically linked.

All these translate to a surge in data center traffic. In its recent Global Cloud Index, Cisco forecasts that global cloud traffic, the fastest growing component of data center traffic, is expected to grow at a 35% combined annual growth rate (CAGR) – from 1.2ZB of annual traffic in 2012 to 5.3ZB by 2017. That will be a whopping 69% of annual data center traffic.

Asia Pacific will experience an even faster cloud traffic growth rate of 43% CAGR. By 2017, the region will generate 1.876ZB of cloud traffic annually and process 36.5 million or 31% of global cloud workloads.

To deliver reliable service and fulfill business objectives, CIOs must quickly recognize and address the impact of increased volumes of traffic and network speeds on critical network and application performance.

Megatrends affecting networks

In 2014, 10GbE switches will outstrip 1GbE ones to comprise more than 50% of shipments and by 2017, 40GbE would comprise close to one-third of all data center switch shipments by 2017, according to Crehan Research.

But most monitoring, analysis, and security tools are incapable of processing the volume of packet traffic at line rate for 10Gb, 40Gb and 100Gb networks, according to officials at Gigamon, a traffic visibility solutions provider. A 10Gb network transports 50 times more traffic than a 1Gb network while the cost to upgrade a 1Gb tool to 10Gb can be multiple times the cost of the existing 1Gb tool.

A majority of these tools are implemented with insufficient visibility to accurately report on activity and secure the data center and network.

Monitoring challenges

The quest for lower cost of IT service delivery and simplified IT operations has pushed up demand for converged infrastructure solutions – defined, repeatable, and scalable solutions.

The challenge here is the use of network virtualization that creates silos of IT as well as a blind spot for monitoring, analysis, and security tools. These tools cannot see inside the virtual switch or overlay networks.

At the same time, server virtualization reduces the capability to monitor and manage many aspects of the data center environment, including application traffic and end-user activity.

Migration to the cloud

Security and compliance are top enterprise concerns in virtualization and cloud deployments. Since tools used to monitor local networks cannot see inside the virtual network, organizations are torn between competing priorities to virtualize the data center and to fulfill requirements for visibility.

They are often frustrated by the lack of visibility despite deploying more and more tools throughout the network and having to manage them. Invariably, traffic visibility, compliance and data security are among the top inhibitors to cloud adoption.

Some organizations may rely on off-premise private clouds or rely on a service provider. The provider may, in turn, outsource its data center operations to a collocation facility, or pool resources among customers but separate them using VPNs. These add to the visibility challenge.

The need to simplify IT

Increased complexity in the data center – with network upgrades, virtualization, encapsulation, overlay networks and cloud solutions – makes monitoring, analyzing and securing IT assets an arduous task. Full compliance is also difficult to achieve because many blind spots remain in the typical data center infrastructure.

An inherent problem in virtualization is the multiple VMs running on multiple servers sharing the resources of a physical server. When one VM takes an excessive share of resources, the performance of other VMs is affected. Greater complication arises when a VM and its applications are moved from one physical machine to another. Enterprises must take these dependencies into account.

Perfect storm

Clearly, effective monitoring strategies and real-time troubleshooting are growing concerns for businesses. As Gigamon officials point out, the race to higher speeds, the emergence of increasingly complex security threats and ruthless compliance requirements are creating a perfect storm for network meltdown.

These challenges indicate the need for a visibility layer solution or centralized fabric that delivers relevant data from various networks under an administrative domain – campus networks, branch/remote networks, private cloud or SDN islands – to a centralized set of tools.

IT administrators who deliver customer-driven service-level objectives need pervasive monitoring and analysis of application and end-user traffic, including traffic in the converged infrastructure and virtualized network.

One solution is an intelligent network monitoring fabric that meets the volume, density and scale requirements of today’s data centers, extending visibility into various silos of IT. Such a fabric should be a critical component of any data center or network infrastructure investment.