Nick Lippis is an enterprise computer networking expert and consultant. He regularly publishes “The Lippis Report” which analyzes current industry trends and interviews strategic vendors. I enjoy his insight and the quality of the content in the newsletter. The most recent Lippis Report, titled “Risk Management Techniques To Increase Network Availability”, dealt with the risk management approach of mixing network equipment vendors in a corporate network.
As both a hedge against consolidation and a risk mitigation strategy executives from large corporations are reviewing the single vs. dual network vendor strategy. Risk management executives striving to increase uptime by mitigating the risk of network outages are recommending a dual vendor strategy. Some business leaders and risk management executives are exploring the concept of mixing network equipment vendors in a corporate network as another degree of redundancy in the hopes of increasing availability even further.
This is a typical thought process for risk-averse management – “don’t put all your eggs in one basket“. But Nick is not supportive of this thought process:
By incorporating a dual network vendor design, the hope is that network outages due to specific vendor equipment faults and/or exploits targeting a vendor’s software will be mitigated by diversifying the number of suppliers in the network. On the surface this approach seems alluring; however, under deeper inspection network availability actually decreases while operational cost increase in mixed vendor environments.
Nick proposes three major reasons why having different network vendors in your corporate network actually lowers your service levels instead of decreasing risk. First, while the goal is to lower costs by having diverse hardware vendors that have different outage probabilities and can be competed against each other for lower purchase prices, having different vendors significantly raises labor costs, which dominate OPEX:
Multiple network equipment suppliers within a network increase network complexity, which drives up operational cost as well as the probability of outages. Complexity inflation increases human capital cost as operational staff is required to be trained and proficient in new management software and its nuances. Complexity inflation is measured by either increased operational hours or additional staff required to manage the network. The end result of a dual vendor strategy is that IT may not be able to operationally support either backbone network appropriately unless an infusion of human capital is appropriated. Operational cost is dominated by human capital cost. Therefore, mixed network vendor environments drive up the most expensive cost component in total cost of ownership (TCO).
Second, while there would be different physical devices from each vendor in the network, these devices are still logically connected via standard protocols. A protocol problem in one vendor’s device could possibly affect the other vendor’s equipment since network devices operate in a multi-protocol mesh:
Network equipment does not operate in isolation as it shares physical and logical connections, which influence system behavior. From a physical connection point of view, the dual vendor network strategy has value in the fact that there are redundant systems; however network devices share information and files plus execute common protocols which they rely upon to perform their basic task of packet forwarding. For example, a malicious attack on a routing table of one vendor would result in corrupting both vendors’ routing tables as routing tables are updated and shared between vendors.
Third, and I feel most importantly, having dual vendors typically relegates network features to the lowest common denominator – standardized protocols. While networking does have the best standard protocols in the IT industry, vendors differentiate themselves by features beyond the standard, for example Cisco’s Flexible Netflow. But with dual vendors, using these enhanced features can be difficult, if not impossible:
But while the networking industry is highly standardized, every vendor in the industry differentiates standard offering with leadership features, capabilities and value add. There is value add to basic standards such as quality of service, congestion management, multicast, network access control, wide area acceleration, application acceleration, route optimization, etc. In a dual backbone architecture deployed with dual network equipment vendors, the services offered by the dual backbone are relegated to the least common denominator of standard offerings. While there may be powerful innovation offered by one of the vendors in the dual backbone of which an IT organization would like to take advantage, it will be precluded from doing so unless those innovations are available on both platforms. Even if the innovation is said to be available from both vendors, the likelihood of it being available simultaneously so that IT can implement and take advantage of it is doubtful as competitors never deliver features and innovation in unison.
I would also add a fourth reason to use a single network hardware vendor – “a single throat to choke”. It’s very frustrating when dealing with outages or new installations that don’t work between two vendors. Vendors, as soon as they find another vendor is involved in the outage, love to play the “it’s the other vendor’s equipment” game. That puts the onus on the customer to coordinate vendors and find the problem instead of relying on the vendors technical support expertise. Nick makes a strong case for a single-vendor strategy. Even service providers, who’s business is building networks, have standardized on a single vendor for their networks to push the envelope of features and services for customers. I have to agree that using a single network hardware vendor is the best strategy, even if the upfront purchase costs (CAPEX) are higher from a single vendor. The long-term costs, both hard (OPEX) and soft (less features, customer unhappiness), can easily be seen. The question is where do the benefits of this single vendor strategy end? I’ll expand on that thought in a later blog.
