We’ve all heard of single-tenant and multi-tenant SaaS applications. What may not be obvious is the economics of the tenancy model. Pundits claim single tenant applications aren’t “SaaS” but that’s not true. Single tenant application can be offered as SaaS however there are cost implications to doing so that should not be ignored.
Quick Review of Single v. Multi Tenancy
A single-tenant SaaS application is an application built upon traditional enterprise application “N” tier architecture. It may support many users (as an Accounting, SCM or HR system might for example) but there is only one instance or ‘tenant’ of the application. There is no concept within the application that there might be more than one instance installed. There’s no supporting structure built by the developers to handle the provisioning and care and feeding associated with multiple instances of the application existing in the same data center.
As a SaaS provider, if you operate single-tenant applications, every time you make a sale you need to chunk down an entire set of the “N” tier architecture associated with the application for the new customer.
A multi-tenant SaaS application is designed and built with the idea that there may be more than one customer using the application instance. This can be accomplished in a number of ways but the end result is that making a new sale does not require deployment of duplicate sets of “N” tier architecture. All other things equal, multi-tenant is the way to go for SaaS providers. Let me show you why this is true.
Economics of Single v. Multi Tenancy
Let’s take a quick look at the economics associated with single and multi-tenant applications. I’m going to use an example constructed around a mythical application using mythical technology. For this example we’ll ignore the network bandwidth costs and make some other simplifying assumptions.
For this exercise let’s assume the following:
Application Type Enterprise Application (SCM, ERP, CRM etc.)
Technology Licensed commercial app server and database
# Tenants in ST App 1
# Tenants in MT App 10
Hardware Cost Assumptions
App Server Price $2,000
Database Price $3,000
App Server # Cores 12
Database # Cores 12
Software Cost Assumptions
App Server Cost per core $3,000
DB Server Cost per core $3,000
Note that the assumption I make in this exercise is that the single tenant application supports only one tenant (thus the name) and the multi-tenant application supports 10 tenants. In the real world the first assumption tends to hold however in my experience multi-tenant applications typically house many more than 10 tenants.
Assume the development costs for the single and multi-tenant application are equivalent then you can see that clearly at any scale the choice is clear. Multi-tenancy is the way to go. You can see this in the table below, which presents the costs associated with scaling our mythical SaaS application up from 1 to 50 customers.
The following simple model shows the economic consequences of hosting single tenancy v. multi-tenancy.
There are some other lessons from this model as follows:
- Single Tenant Apps have more infrastructure, therefore more labor
- The economics of scale are flat or negative for the single tenant app
- The economies of scale are positive for the multi-tenant app
- The economies of scale are directly related to the # of tenants per instance
This model grossly over simplifies the situation in many regards but the lesson is clear: Single tenant applications don’t scale up in terms of costs. Go with a single tenant design and you’ll need an army of people for care and feeding of the application.
Deploy a multi-tenant application and you’ll reap the benefits of economies of scale.
Want to learn more about the latest changes in IT? Download our extensive e-book: Defining Modern IT.