Network Infrastructure Upgrades: Driven By More Than Just Bandwidth

Hoping to improve demand for products and increase sales, networking vendors are taking a multifaceted approach to driving enterprises toward large-scale network infrastructure upgrades. Users should evaluate potential drivers in the context of one another to determine whether the net effect is sufficient to warrant an overall upgrade.
Written by Chris Kozup, Contributor on

Hoping to improve demand for products and increase sales, networking vendors are taking a multifaceted approach to driving enterprises toward large-scale network infrastructure upgrades. Users should evaluate potential drivers in the context of one another to determine whether the net effect is sufficient to warrant an overall upgrade.

META Trend: Enterprises will renew spending on campus LAN initiatives in 2004, with new allocations to voice and data convergence, wireless LANs, network security, and high availability. Large enterprises will further consolidate spending with one or two strategic providers. By YE05, Gigabit Ethernet to the desktop will become the preferred choice as pricing nears that of Fast Ethernet. Wireless LAN security and quality-of-service standards will be ratified, shifting the emphasis to management across wired and wireless domains.

Not since the major upgrades to switched Ethernet and hardware-based routing during 1997-01 has the IT organization (ITO) had such a broad requirement to overhaul the network. At the time, enterprises were in need of additional scalability within the network core, necessitating a migration to Gigabit Ethernet in the backbone. In addition, most enterprise networks evolved into a mesh of independent islands, each running its own protocol and having its own characteristics. Consequently, ITOs upgraded their infrastructures to gain capacity as well as streamline and consolidate services and protocols to improve manageability.

The dominance of Ethernet, stability of standards, and presence of modular networking platforms capable of supporting traffic flows up to 10Gbps mean that enterprises can take a more gradual approach to upgrading their network components. META Group has noted the trend for enterprises to extend the life cycle of their network infrastructures well beyond the three-year depreciation cycles, frequently into the six- to seven-year time frame. This trend was exacerbated by the poor economic environment and tendency for capital investments to be denied or severely stunted. There are many small reasons to consider infrastructure upgrades, each of which on its own is not sufficient to push users to change. However, in combination, they present a decent argument for overhauling networks that are more than three years old.

Upgrades Are Not Driven By Capacity Alone. Through 2004, enterprise infrastructure spending will grow at a modest rate, with many ITOs refocusing on neglected campus network investments. However, even with this improved spending environment, few enterprises have the need to upgrade the network for the traditional reasons of decreasing contention and increasing capacity. Despite the proliferation of applications, most enterprises have provisioned adequate bandwidth to the desktop and in the core. Average utilization rates of desktop Fast Ethernet ports remain below 10%-15%. Networking vendors are wise to the lack of bandwidth drivers to upgrade and will shift their marketing focus toward the softer benefits of upgrading during 2004/05. Such benefits include security, management, quality of service (QoS), convergence readiness, wireless integration, and high availability. Longer term (2008), the enterprise network evolves into a truly multiservice infrastructure capable of supporting the breadth of business applications in a highly predictable, secure, and resilient fashion.

Security. IT and business executives have become acutely aware of the need for additional security across all aspects of the business. The network is no exception. Leading networking vendors have integrated security features directly into the thread of the network, often blurring traditional boundaries between security operations and network operations, though leading vendors must focus attention on separating the management views of these two constituencies. Appliance-based security technologies (e.g., firewalls, IDS, VPN) are now routinely being integrated into switching and routing products. Although not every organization will choose this approach, many security enhancements made to switch and router software will lead enterprises to consider upgrading existing products. Features such as IEEE 802.1X port-based authentication, secure shell (SSH), authenticated VLANs, and SNMP Version 3 are all examples of security enhancements to network infrastructure that can improve the overall level of port-based and network security. Some META Group clients have upgraded older products to take advantage of more enhanced security features, though security and networking domains remain largely separate due to operational requirements.

Product Obsolescence. Nothing will cause an ITO to migrate products like the imminent end-of-life of that product or the vendor’s exit from the market. Although the majority of networking vendors selling into the enterprise market are relatively stable (having made it through the worst of the economic downturn), many are in the midst of product transitions. As a primary example, during the past 12 months, Cisco announced the end-of-life of several Cisco router models (some versions of Cisco 2500, 2600, 3600, and 7200), and some of its low-end switches, including Catalyst 2900xl and 3500xl. Other vendors such as Extreme Networks, Foundry Networks, Enterasys, and Alcatel have also introduced newer products to supersede their older generations. Users should always push vendors for details on expected product life and negotiate trade-in values for such products that will be replaced in the near term (e.g., 12-18 months).

Convergence. Most organizations are planning to adopt a converged communications infrastructure by 2007, using the IP network as a platform (see Delta 2680). To guarantee the appropriate level of responsiveness for mission-critical real-time applications such as voice and video, networks must be capable of differentiating and prioritizing packets in a way that is simple to manage and maintain. The resulting need for multiple QoS levels will more often than not require an upgrade of wiring-closet switches to support new Layer 3 functionality. In addition to QoS, power over Ethernet has emerged as another beneficial feature of next-generation products (see Delta 2482). The current price premium for powered switch ports is 20%-50% greater than standard Ethernet. However, component advancements will decrease this pricing to 10%-20% premium during 2004/05. At this point, it will be cheaper to purchase such features rather than retrofit in future years.

Wireless Integration. The enterprise interest in Wi-Fi is forcing an architectural shift toward integrating wireless features into the Ethernet switching infrastructure. Switching and routing infrastructures will increasingly enhance the reliability, security, and services delivered by the Wi-Fi infrastructure. ITOs must evaluate Wi-Fi integration in the context of their existing data networks.

High Availability. High-availability networking is an additional driver for network upgrades. Enterprises have become increasingly aware of the importance of just-in-time access to corporate resources and services. Voice services need more reliability than the typical LAN provides. Server consolidation increases the need for network uptime. On demand data center services also assume a highly reliable network. Without the network, applications are not delivered. Communication is hindered, and business grinds to a halt. Many vendors (e.g., Alcatel, Cisco, Extreme, Nortel) have made significant investments in both component-level reliability (e.g., hitless management failover, redundant power supplies) and system-level reliability (e.g., VRRP, improved spanning tree). Additional improvements in software stability and network design will also add to the level of overall network availability.

Management. Network management must span multiple product categories, extending beyond simple fault management into configuration, performance monitoring, and application discovery. In fact, network management should ideally integrate into an overall enterprise management architecture. A management approach that takes a system-level view into features and allows partitioning of views will be critical as the network evolves into a single infrastructure supporting multiple applications (e.g., data, voice, security, SLA view).

Price Improvements. Decreases in component costs have enabled vendors to offer additional feature/functionality at similar or lower prices than was the case in older products. This is a key factor when weighed against a potential rise in maintenance pricing and support issues for older equipment. A primary example of this is the prevalence of Layer 3 functionality in both modular and fixed wiring-closet switches. By 2005, thanks to declines in chip prices, vendors will be able to offer 10/100/1000 Ethernet ports for the same price as current 10/100 Ethernet ports. This “more for your money” proposition is often the reason enterprises buy into features that otherwise are not currently being employed.

Bottom Line: Bandwidth is no longer a strong driver effecting campus network upgrades. Users should consider requirements in areas such as security, high availability, convergence, maintenance, product life cycle, and investment protection when evaluating future networking infrastructure upgrades.

Business Impact: A well-planned procurement strategy will enable the business to optimize its spend and receive the most "bang for the buck."

META Group originally published this article on 11 March 2004.

Editorial standards