Should a Managed Service Generating MRR Qualify as Cloud?
One significant point that is often overlooked in the cloud/MRR discussion is that most of the traditional large enterprise telephony solutions are single instance. This includes offers from Avaya, Cisco, Mitel, and NEC. Most analysts and industry pundits (and multitenant UCaaS vendors) argue these aren’t cloud, even if they’re deployed in an MRR business model.
In a single-instance model, dedicated software “instances” are deployed for each customer organization, which generally makes offering dedicated-instance cloud solutions cost-prohibitive for small-scale installations. In the SMB UCaaS delivery model, a multitenant system is critical to delivering a competitively priced cloud solution. However, while this pedantic view of cloud as requiring multitenancy is valid for smaller organization deployments, for larger organizations, the cost of a dedicated single-instance decreases with scale and eventually becomes irrelevant.
For example, a typical Avaya Aura or Cisco Call Manger solution has a minimum of four to six servers or virtual machines (VMs) for a redundant, highly available deployment, and may have dedicated gateway hardware as well. This number grows as additional services and capabilities are added. In a single-instance solution, servers, networking, and other resources are dedicated to the single customer, regardless of scale.
In our experience, the five-year cost for the dedicated infrastructure (servers, networking, data center, support, etc.) is around $50,000, which when amortized over five years has an annual cost of approximately $10,000 per year for the UCaaS provider to offer this type of solution. If this was deployed for a five-person company, the cost to the provider for each user would be $167 monthly just to cover the costs of the “core” deployment. In a multitenant model, the five users share the core with tens of thousands of other users from other organizations, so the cost per user of the core is a small fraction.
The figure below compares a UCaaS provider’s per-user cost and the organizational cost for the multitenant and dedicated single-instance models. The chart shows the three cost components of service delivery (typically called cost of sales in finance terms).
- Per-User Costs: The green is the cost of providing the service to the individual user. This includes all the real-time processing, the individual services (telephony, conferencing, collaboration, etc.), bandwidth, PSTN access costs, most technical support, and anything else that scales linearly with the number of users. Generally, this is 90% to 95% of the cost of service delivery, as it reflects most of the real operational costs over the life of the solutions. This includes all the infrastructure resources required to support the individual user.
- Per-Company/Customer Costs: The blue area represents costs associated with the company/customer. These include onboarding, billing, company level support, etc. Typically, the costs of supporting an organization are relatively small.
- Single-Instance Infrastructure: The orange block on the Single-Instance chart (at right) shows the amortized cost per end user for the dedicated equipment required for a single-instance deployment. Compared to a multitenant solution that may support 100,000 users, a dedicated instance for 500 users on the same level infrastructure will have a large unused capacity. This is reflected in the higher-per-user costs when the number of users per organization is small.
If the cloud services are essentially the same, the only real cost difference between multitenant and single-instance deployment is the cost of the dedicated infrastructure required for the latter. As can be seen, per-user and per-company costs for each cloud seat are the same except for the infrastructure required for a single-instance deployment.
On the multitenant systems on the left, you can see how this impacts the delivery costs of a multitenant base offer. As company size grows, the actual decrease is minimal given the relatively low per-company costs. The right chart shows the same analysis for a single instance. As you can see, for deployments with fewer than 100 seats, the cost impact of the dedicated equipment makes a single-instance uncompetitive. This can be seen from the red line at the top showing comparative costs at 100 seats. However, at about 1,000 seats (the red line in the middle), the added cost of the dedicated infrastructure for single instance is now allocated over a larger population and the added cost becomes insignificant.
Understanding this cost structure illustrates why virtually all the Cisco HCS providers have a 400-seat minimum for an HCS deployment to the cloud. However, as you move to 1,000 seats and above, the added allocated cost per user of the single-instance deployment drops dramatically. By this point, the $10,000-per-year dedicated core costs, when applied to 10,000 users, becomes only eight cents per month. It essentially becomes irrelevant in the overall UCaaS cost to deliver a seat. So, a single-instance deployment becomes very competitive at 1,000 seats and above. At the point the capacity of a unit of softswitch computing is equal to the size of a single company UCaaS deployment, there is no difference at all.
Continued on next page: Conclusion
