Specific numbers describing how many software companies (publishers) have made the move to the SaaS model can be hard to come by. A Gartner report(1) indicates the shift, on-premise to cloud-based, is growing at a rate of 11.5% (2021). Traditional on-premise software is shrinking about 1.6% at the same time.
In part, I believe the disparity comes from most new software originating in the cloud now. And, many companies prefer to create new-for-the-cloud software over substantially rewriting an existing tool.
In the late-90s, the internet made possible the rise of the ASP (application service provider). The ASP approach basically launched a remote server (with its own OS, Software, and hardware requirements) for each subscribing customer. Unfortunately it proved unscalable and, with low bandwidth then (1Mbps was huge), the ASP model experienced a quick death. (In short, the experience sucked.)
SaaS, emerging around 2000, gained traction by the late 2000s as a new model for delivering software apps over the internet. By 2010, SaaS had come into its own and is the dominant model today.
SaaS (Software as a Service), pronounced ‘sass,’ can refer to either the licensing mechanism (i.e. Subscription) or the deployment model (i.e. on-premise or cloud-based). For this post the focus is primarily the deployment model.
For established software companies the conversion from on-premise to cloud-based can be quite challenging. In 2022, I was part of the team assessing how to move our own product to the cloud. There were 5 core drivers:
- Scalability – Cloud-based software is more scalable with access to virtually unlimited resources.
- Cost Savings – Multiple considerations but, for the customer, can mean improved pricing options. Greater flexibility for customers with on-demand requirements.
- Flexibility – Access the software from any device with an internet connection.
- Security – Address existing licensing concerns while also making updates easier and more secure.
- Competitive advantages – Meet customer expectations we currently couldn’t. Deliver on those same expectations where our competitors were lacking.
Perhaps most constraining for us were the resources and time to do a substantial rewrite of our existing tool, optimized for the cloud. This mitigated many of the scalability benefits. It also impacted several of the other core drivers.
One of the approaches we evaluated was offered by Oracle. We discovered they had essentially brought the ASP model (reason I described it above) back to life.
Rather than rewrite the code to support a multi-tenant implementation (e.g. multiple users on single instance), users could subscribe to our software, hosted on an Oracle server. On demand, Oracle’s system would spin up a virtual PC, complete with an installation image (OS and our software) for their use.
Conceptually, the process is simple. We, the software publisher, specify the hardware requirements to run our software, including OS, CPU, RAM, storage, if a GPU was required, and so on. The solution proved technically feasible albeit with some graphic anomalies during testing.
Going beyond the ‘will our software work’ testing, we discovered Oracle’s solution only provided the backend (run the software) component. The reality was we would need to build a complete front end for the solution. This would need to include ecommerce for all account and billing administration. We would also need to initiate, monitor, and terminate each customer instance. Beyond needing to seek out skills or solution components we did not have, this brought to light two major concerns.
Modern-day internet and hardware performance has made some versions of the old ASP model possible again.
One, was what would happen if we did not have an absolutely fool-proof instance management engine? During proof of concept testing, one of our evaluators left their instance live for several days, not realizing the need to specifically terminate the instance. The bill was several thousand dollars.
Another, was simply cost. During proof of concept, we realized we would need to use specify higher performance virtual machines compared to what a customer would normally need on premise. Combined with the hard-learned cost structure, it was clear this approach would not let us offer a cloud-based solution at an acceptable price point for our customers–without considering costs for the required front-end development.
Summary
The shift in software delivery models from on-premise to cloud-based continues, with SaaS being the dominant model today. However, the conversion from on-premise to cloud-based can be challenging for established software companies.
We focused on five core drivers for making the move to the cloud: scalability, cost savings, flexibility, security, and competitive advantages. Resource constrained, we had hopes the revived ASP model offered by Oracle would be a viable solution. Ultimately, we found the expense and delivery model to be prohibitive.
While the shift to cloud-based software is growing, it can be difficult and costly for some companies to make the transition.
(1) Gartner. (2021). Forecast Analysis: Enterprise Application Software, Worldwide, 1Q21 Update. Published on 4 June 2021. Gartner subscription required to access the report.
Image Credits:
Software code – Markus Spiske
Servers – Matthieu Beaumont