The concept of “as-a-service” or pay-per-use has become pervasive across the IT industry. The value proposition of as a service is that businesses can pay for things as a monthly subscription over a long period of time instead of having to fork over huge sums of money now. Traditional procurement methods create incredibly lumpy spending patterns as a one-time purchase is made at year zero and then minimal spend is made over the next several years until the next upgrade when the cycle starts again.
Most technology markets have already made this shift as IT departments can buy software, compute processing power and storage in an as-a-service model. However, one market that has yet to evolve is the networking industry. In addition to having big spikes in spending, the traditional purchasing model has other complexities.
Typically, businesses must overbuy the number of ports they actually need for future scalability. For example, there may only be 10 people and a few other connected devices in a department or branch office. However, businesses purchase a 48-port switch to future proof the deployment because adding more capacity requires buying new hardware. That means the organizations are paying for a number of ports that may never get used. Even if they do use all the ports, it’s likely they would buy a second switch to keep some room for expansion. Extrapolate this across the company and it’s easy to see why businesses are paying for a significant number of unused ports across the company.
Also, when network infrastructure is managed on a box-by-box basis, it’s often purchased that way. This leads to complexity in ordering and managing licenses. IT organizations typically buy, manage and upgrade hundreds or even thousands of separate software licenses. This is one reason why ZK Research (I am an employee of ZK Research) estimates that large organizations run more than 30 different software images across the network making ongoing management extremely challenging. Also, the wide variety of network software features are typically sold a la carte, meaning customers must determine the right set of features for every point in the network. This can lead to inconsistent features across the network, which can be particularly problematic when implementing features network wide to optimize application traffic and secure the network.
A true network-as-a-service offering would simplify the procurement and management of network equipment. First, it would shift the purchasing to something that is more consumption based. This could be done on a per-port basis or even on network utilization. However it’s done the benefit is that organizations would pay for only what they are using and then expand the deployment as needed. This alone could save a significant amount of money over traditional purchasing methods.
Also, network spending shifts to an OpEx model where the company pays a “per month” fee for the network. This creates a flatter, most consistent spending pattern that is easier to budget for. The pay per use model also leads to more consistent upgrades. Instead of trying to squeeze every last packet out of a switch, businesses could upgrade the product at every contract cycle ensuring the company always has current technology. Also, if network upgrade cycles are consistent, this prevents the problem of having inconsistent software and features across the network.
A few vendors have tried to bring the concept of “as-a-service” to the network, most notably Brocade (Brocade is a client of ZK Research). The company currently has a network subscription service that’s close but the company tends to use it primarily in certain verticals, such as government, and it’s not a true as-as service as it has a 60 day out, but it’s the closest thing the industry currently has.
Also, Cisco ONE (Cisco is a client of ZK Research) has many economic benefits for customers but still has an upfront hardware component so it’s not a true “as a service” offering.
Given the increasing shift for IT to move to an as-a-service model, it’s time for the network industry to make this shift as well.