The Disaggregation of the Enterprise Network
Market competition, innovation and a renewed focus on business requirements have led to a resurgence of multivendor network architectures. This renewed approach offers improved functionality and lower costs than a single-vendor architecture, with little or no additional operational complexity.
The enterprise networking market has migrated from a single-vendor environment to one in which discrete decisions for different network building blocks are rapidly becoming the norm. For network architects, this new approach will lead to network solutions that better meet business requirements and save money, without adding operational complexity.
- No network vendor has "good enough" capabilities across all the network functions required for a distributed large enterprise.
- Organizations are rapidly embracing a multivendor, best-of-breed approach to building enterprise networks.
- Stop making assumptions concerning future network architectures, technologies and vendor relationships.
- When starting projects to upgrade network components, focus on business requirements.
- Never award default business to incumbent vendors (especially in adjacent market areas) without a review of changing business requirements and financial relationships.
The network infrastructure market has evolved rapidly during the past three years, from one in which most organizations adhered to a single-vendor architecture to a more business-driven network architecture that increasingly uses a multivendor, best-of-breed approach. Our research found there is no operational advantage to maintaining a single-vendor approach (see "Debunking the Myth of the Single Vendor Network"). That leaves analyzing the functional and financial differences between vendors as the key requirements for building a network that best fits business requirements.
Although there is no operational justification for choosing a single-vendor architecture, we believe that enterprises would benefit from evaluating the vendors' operational capabilities, because there is evidence that different approaches to network equipment design and operational focus can reduce network complexity. Those that follow this new approach will find that they build a better network without functional compromises or unnecessary overbuilding, while saving substantial amounts of their total network infrastructure budgets — in most cases, demonstrating 20% to 25% total life cycle savings. Functional and financial analysis has replaced sole-sourcing convenience, due to the compelling benefits of a disaggregated approach to network design.
The first years of the 2000s saw a strong movement toward network integration and the prevailing thought that a single-vendor network would simplify operations and provide a better, more predictable network experience that served users well. Although a "one-stop shop" simplifies procurement and creates strong ties between the network operations staff and its supplier, operational benefits did not accrue from this approach. As we described in "Debunking the Myth of the Single Vendor Network," there was no operational benefit from relying on a single-vendor environment for all networking requirements.
The reasons for this lack of operational leverage are twofold. The first is that the network is not a homogeneous mass of technology, but, rather, a set of related technologies and systems, each with their own functions and requirements. Although integrating functions, correlating faults, sharing data between management systems, and creating common policies and other integration tools is possible, it's not a simple task.
The second problem is that incumbent network vendors have not typically focused on operations, and we've seen little progress from the larger network equipment players in the market to solve these difficult problems. This can be due to vendors growing via acquisition, and, therefore, having products with different management capabilities and approaches or vendors that lack portfolios broad enough to deal adequately with broader network operations issues.
Despite the shortcomings in the market, the shift toward network disaggregation needed some key market conditions to get networking organizations to consider the shift back toward a multivendor, best-of-breed approach.
- Significant product differentiation — functional or financial
- Viable mainstream competitors in key markets
- Stronger incentives for buyers to change behavior
The past few years have seen a confluence of events that completely changed the enterprise market dynamics and shifted the market toward a disaggregated network architecture:
- The emergence of HP in 2006 and, later, Juniper in 2008 as viable, mainstream, top-tier competitors in L2/L3 switching and routing markets
- The economic downturn of late 2008, which continues to encourage enterprises to re-evaluate their buying practices
- The increasing importance of markets in which functional differentiation is key to meeting enterprise requirements, such as application delivery controllers (ADCs), WLANs, WAN optimization controllers (WOCs), data center fabrics and various aspects of network security
Enterprises now have significant buying power. Adopting newfound best practices and a growing realization that it really is possible to build a better network, while lowering costs, creates an incentive for buyers to change behavior. We believe there are three clear-cut reasons for disaggregating the network. From analyzing market behavior, we see that all three are having an increasing impact on architectural and procurement decisions.
The primary reason that the market is moving toward a disaggregated network architecture is functionality. Although parts of the network architecture are moving down the commoditization curve, where good enough is perfectly fine (workgroup switching is a prime example), other areas, such as wireless LANs, WAN optimization, ADCs and data center fabrics, clearly represent innovation in which there is meaningful differentiation among vendor solutions. These noticeable differences among vendor approaches can provide real business benefits. Choosing the wrong solution in these areas can compromise business processes and affect the success of the business. Looking across the broad network requirements, it's clear that no single-vendor network is good enough for each of the required network building-block categories.
Strategic Planning Assumption: Through 2016, no single-vendor solution will offer the required level of functionality for a distributed, large-enterprise network architecture.
The economic downturn of 2008 that affected most global economies forced many organizations to scrutinize their IT investments more closely. For network equipment, this resulted in the discovery that the delta between their incumbent vendor's price and what they could achieve in a competitive environment had grown to unanticipated and unprecedented levels. Today, it's possible to see capital cost savings of 50%, even when buying the same equipment from the same vendor. Further equipment and maintenance savings are available when looking at alternative vendors. By considering a multivendor architecture, most enterprises can reduce their capital budgets by at least 30%, with a potential reduction in their total network equipment and operations costs of nearly 25%.
As we documented in "Debunking the Myth of the Single Vendor Network," there is no operational justification for maintaining a rigid single-vendor architecture. This goes against the common wisdom that reducing network infrastructure to fewer vendors (or just one) results in reduced operational costs. To find the truth in that conventional wisdom, we have to look deeper. It's not just simplifying to fewer vendors that's required, it's reducing the number of device models and OS versions. Economies of scale and operational efficiencies can be gained by establishing "gold standards" for configuration and performance policies; however, this can only be effective once the number of device models and OS versions in the network architecture has been reduced. Our research has shown that, in many cases, adding a network vendor for one or more of the building blocks actually had the counterintuitive effect of reducing the number of device models and OS versions that needed to be managed.
Most network operational tools in the fault and performance management categories already use industry-standard, vendor-independent Single Network Management Protocol (SNMP) to interface with and gather status data from network devices. Introducing a second network infrastructure vendor doesn't create any operational complications, because these same management tools can continue to be used.
Our client inquiries have shown a growing adoption of multivendor network architectures and a move toward a more disaggregated approach to network design and procurement. This is clear from the types and volumes of questions we are getting:
- We now receive few client inquiries that assume the LAN switching vendor should be selected for all LAN switching requirements from the workgroup edge through the data center core. In their place, we now get discrete calls about data center networking, workgroup switching and campus backbone architectures. The requirements for LAN switching have been disaggregated into their discrete parts.
- It was once common to review bids that included many different product families — for example, campus and data center switching combined with ADCs, firewalls, routers, network access control and wireless LANs. Most of these are now separate decisions. Rather than giving one vendor default business across many areas, these deals have become separate building-block decisions. For example, the campus network becomes competitive between Cisco and HP, the data center among Cisco, Juniper and Brocade, and the ADCs between F5 and Citrix.
- The volume of client inquiries requesting advice on what shortlist vendors are appropriate and pricing reviews of competitive network equipment bids has jumped approximately threefold since 2008. Functional and financial analyses have replaced sole-sourcing convenience, due to the compelling benefits of a disaggregated approach to network design.
Gartner's primary recommendation is to stop making assumptions about architectural directions, technologies, vendor relationships and products. Don't simply award default business to an incumbent vendor, especially in adjacent areas. Make your vendors earn your business each time you go through network upgrades and refreshes. This doesn't imply that you shouldn't continue to buy from your incumbent, but ensure that you're in control of the process and that you're making decisions in the best interests of your organization. Finally, treat the network as a collection of building blocks. The market is moving toward a disaggregated, best-of-breed approach to take advantage of the considerable functional, operational and financial benefits.
Through 2016, no single-vendor solution will offer the required level of functionality for a distributed, large-enterprise network architecture.