[The] lack of skilled personnel is what makes IaaS or PaaS more trouble than they're worth.
Cloud computing is an economy of scale. Enterprises can get a better deal in the cloud if the cost of IT infrastructure for a provider, at public cloud scale, is lower than internal enterprise IT costs. The question is: What are the components of "cost"?
Lower-level cloud services like IaaS essentially involve hosting virtual machines (VMs), which can save the user hardware costs in dollars as well as power and cooling, but displaces little to no software costs. PaaS reduces both software costs and software support costs. SaaS, on the other hand, provides the greatest cost reductions -- from hardware up to the application layer -- and it has the biggest affect on support and operations costs. In general, SaaS presents a better economy of scale.
Where SaaS falls short, though, is that it's hardly a model that can address all enterprise IT costs. SaaS is suitable for packaged software, not custom applications. No enterprise will cloudsource its core mission-critical software. This has pushed SaaS out as the universal cloud model, forcing early cloud adopters to seek a different approach that could meet all their needs. IaaS and PaaS fit that bill.
Second thoughts on SaaS
But the tides may be turning again. Experience in the cloud and improved dialog with senior management have enterprises taking a second look at SaaS for two reasons:
- Even though enterprises don't expect to cloudsource the mission-critical component of IT, they will look to SaaS for backup or overflow workloads. That means that nearly all enterprise cloud adoption will be in the form of a hybrid cloud with some internal IT services.
- Enterprises are discovering more cloud opportunities on the fringes of mission-critical apps, particularly in data mining and business intelligence. These applications are a good fit for SaaS.
In fact, enterprises state that much of their IT growth has come from "fringe" applications that surround core business applications and exploit data. Unlike core applications, fringe apps are suitable for SaaS. And because they often aren't fully implemented on internal IT sources, fringe apps also allow enterprises to quickly realize incremental savings.
Cloud providers also see the benefit of such applications. IBM has made major initiatives in cloud BI applications; network operators like AT&T and Verizon have announced partnerships with software vendors for cloud hosting of application software, creating a two-tier SaaS product.
SaaS finds its niche
SaaS isn't the right cloud model for all businesses. Small- and mid-sized businesses are typically more interested because SaaS takes the largest bite out of technology support costs. It's ideal for companies who can't afford a large IT staff or who can't retain skilled personnel.
And that lack of skilled personnel is what makes IaaS or PaaS more trouble than they're worth. If a company adopts IaaS to displace an internal server farm generally, it quickly discovers that it's just as complicated to maintain software images in the cloud as it was to maintain software in their own data center -- not to mention the complexity involved in cloud management.
Some software design and deployment trends also favor a SaaS-based service set. Most enterprises have adopted a service-oriented architecture (SOA) model for software, and nearly all software packages created today are SOA-enabled. This means that enterprises can mix custom software components with packaged ones. Offering packaged software components as SaaS facilitates cloudsourcing access to software, either full-time or during periods when internal IT infrastructure is congested or in a state of full or partial failure. And because SOA principles also form the foundation for PaaS, this trend in software evolution may gradually blur the distinction between PaaS and SaaS.
The biggest apprehension with SaaS had been data security and performance, but progress is being made in these areas as well. Where enterprises adopt a Database as a Service (DBaaS) model, for example, SQL Server queries made against applications are sent to database servers. This, in turn, makes the application independent from disk access.
The enterprise -- with proper security precautions -- then can make internal databases available to its cloud SaaS applications. Only the results of an SQL Server query need be exchanged with the cloud application in this case, not the complete parsing of the enterprise database. The growing interest of cloud providers like Salesforce.com in DBaaS further supports this trend and creates a specific model of secure data in the cloud that enterprises can adopt confidently.
ABOUT THE AUTHOR:
Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982.
This was first published in July 2011