Much has been written about cloud, big data and analytics, mobile, social, and IT security as significant forces that are transforming information technology and business. Much less has been said about the impact of these forces on networking— or about the role that networks play when business initiatives are based on them. As these technologies move from future trend to mainstream, the network as the common, critical infrastructure component will get close to the breaking point.
How close is that breaking point? The pace of adoption is telling. Two years ago, only about half of the respondents to the IBM Business Tech Trends Study had adopted analytics and mobile. Social and cloud were deployed by even fewer (34 percent and 39 percent respectively).
Fast-forward to today, and adoption is the norm. Each of the four technologies has now been deployed by at least 70 percent of enterprises—and significant deployments of cloud and social have almost tripled. What’s more, three out of four organizations plan to ratchet up their investments in big data and analytics, cloud and mobile. Yet according to Gartner, through 2017, 25 percent of big data implementations will fail to deliver business value resulting from performance problems due to inadequate network infrastructure.
What many of these organizations have in common is an enterprise network designed a decade or more ago for a very different environment from today’s. Geographic reach, volume and type of traffic, number and kinds of end points, and location of users all tended to be static or relatively slow to grow and change. Data center technology changes such as server and storage virtualization as well as new enterprise network technologies such as virtualization and software defined networks are “bolted on” as needed, adding everincreasing complexity to the network architecture and management.
The network’s role in cloud
Of all the technology forces shaping information technology, none is more defining than cloud. Today cloud has become a mainstream component of enterprise architecture. By making infrastructure, platforms and applications available as a service, cloud has forever changed the way IT resources are delivered and consumed. It has since become a powerful business enabler for mobile, social, analytics and innovation.
For clouds to deliver on their full potential, they must achieve the best possible use of all available resources—processing power, memory, storage and the network. The network plays an essential role in how efficiently the other IT resources are connected, utilized and secured. The network is also the critical connector between cloud solutions and traditional IT components, wherever they are located in the enterprise.
Because of its role in the cloud architecture, the network should be addressed in the early stages of cloud consideration as an essential element in cloud design and implementation— whether the solution is for a public, private off-premises,private on-premises or hybrid cloud. Each of these models poses different considerations and challenges to network design.
Connectivity to the cloud
For cost efficiency, the Internet is a common networking option for connecting to clouds hosted by a third-party provider. This immediately raises the question of IP addressing. Can you use your own IP addresses, or do you need to use IP addresses provided with the cloud service?
More important are questions of security and performance. When employees use the Internet to access enterprise cloud applications in off-premises locations, network design and enforcement of security and privacy policies must encompass the Internet and the public domain. For many organizations, an Internet virtual private network (VPN) offers the protection required. VPNs can also be employed for connectivity between the data center and between multiple cloud instances. VPN services may be offered by the cloud provider at an additional cost or by the enterprise. Either way, careful planning is crucial. Another option is to extend the enterprise network with a secure dedicated connection to the cloud service provider.
Finally, the architectures of the applications implemented in the cloud can also impact their performance over the network. For example, analytics applications for big data will generate large but sporadic bursts of data traffic, while a mobile application on the cloud might see low but consistent volumes of data with higher I/O activity. The use of flash technology as a storage device or cache in cloud architectures requires a low-latency and high-bandwidth network to keep pace with access demands.
If applications use voice and video then the network needs to be designed to handle that real-time traffic with the appropriate quality of service. It is important to understand what the cloud infrastructure is able to offer and design the network access accordingly. Therefore the network requirements for each cloud application must be considered in order to achieve the target quality of service, security and performance requirements for those applications.
Connectivity within the cloud
To deliver on the benefits of cloud computing, applications, servers, storage and the network must be considered as a system and managed and provisioned jointly for optimal function. This requires a new approach for the network— and can be a critical factor in the performance of cloud applications. For cloud providers, employing network virtualization overlays allows a single physical cloud network to be shared by multiple clients while providing the isolation necessary to meet client security and business requirements. The benefits are similar to those of compute virtualization. For example, network virtualization overlay into the virtual instances enables multiple tenants to share a single, physical data center network, providing higher network resource (switch) utilization and lower client cost.
The importance of network infrastructure to big data and analytics
The focus on storing, managing and processing big data and using analytics to extract business value has somewhat overshadowed the importance of networks in the overall equation. This may be in part because line-of-business executives tend to be more involved in the big data discussion than in other areas of technology. But if the data can’t get to the right place at the right time, other infrastructure issues are moot.
As noted earlier in this paper, big data and analytics are moving into the mainstream as pilot projects make way for enterprise initiatives. And there is no doubt about the scope of “big” in big data. Almost half of 540 IT decision makers polled by QuinStreet say that in a typical month, they are already managing 10 TB or more of data for analytics—and 21 percent are managing 100 TB or more. That’s why 41 percent say increasing network bandwidth is a top priority in preparing infrastructure for big data.
Latency is another issue, particularly for analytics applications such as recommendation engines that are serving up real-time results to consumers. Analytics applications continue to get more sophisticated and are using data from an ever increasing range of sources to synthesize their recommendations. This means that a careful analysis needs to be done of the locations of the data being used and the associated data center and storage area network capacities. When possible data should be moved closer to the analytics engines and the engines distributed closer to the users, with appropriate network access capacity designed to meet the response times.
Mobile in the enterprise is becoming the norm
It goes without saying that networks are the critical supporting infrastructure for mobile initiatives. Traditional network architectures will quickly become oversubscribed as the volume, variety and velocity of data change over time. Last year, mobile data traffic was almost 18 times the amount of traffic across the whole Internet a little over a decade ago, and half of it was mobile video.
Wireless LANs go from convenience to business critical
For “bring your own device” (BYOD) and other enterprise mobile initiatives, mobile users will increase traffic that flows across both WAN and LAN infrastructures as they access corporate applications via smartphones and tablets. WiFi, or wireless LAN (WLAN), has become the primary connectivity method for user devices in the workplace, offloading traffic from cellular networks. With the increasing number of mobile devices used for business, demand will overwhelm current WLANs, which were installed as conveniences—not as business-critical networks. By design, the performance,security and management capabilities of these WLANS will typically prove unable to meet the demands of a surge in mobile devices that come with built-in, always-on WLANaccess capabilities.
Social business adds strain on networks
For competitive advantage, organizations are increasingly employing social applications to reach and respond to customers and increase customer intimacy. Most of these social interactions occur in real time and support a wide range of bandwidth-intensive applications and technologies (analytics, wikis, video conferencing, video streaming, social networking and so on). They compound the strain on enterprise networks because they have to share bandwidth with mobile enterprise users and traditional business applications.
New perils threaten network and data security
The pervasive use of personal mobile devices in the workplace dramatically increases the risk to networks and data privacy— as does the increase in business conducted over third-party networks using third-party applications. By 2018, the percentage of off-network corporate data traffic will grow to approximately 25%. It will be driven by the growing adoption of mobile devices and SaaS applications. Eighty percent of employees will use mobile devices.
Put network and security on the same team
Task one is to find the right balance between network performance, accessibility and security. This can only be accomplished by having network designers—who are focused on performance and network access—work in tandem with security specialists whose design criteria are driven by minimizing points of vulnerability to attack. This becomes even more important for the data center, where overlay technologies are enabling a firewall model that distributes security rules across all the virtual switches used by virtual machines and containers to access the physical network.