Cloud Application versus On Premise, Myths and Realities

August 6, 2010
104 Views

I was reading something recently on SaaS TCO that once again dredged up one of our industry folktales about SaaS versus on premise software and it reminded me that no matter how mainstream I think the SaaS apps market has become, there’s still a lot of misunderstandings and misinformation rattling around with prospects, analysts and vendors.

I was reading something recently on SaaS TCO that once again dredged up one of our industry folktales about SaaS versus on premise software and it reminded me that no matter how mainstream I think the SaaS apps market has become, there’s still a lot of misunderstandings and misinformation rattling around with prospects, analysts and vendors. My first reaction was one of frustration that we still have some “experts” pushing concepts that I would have thought we put down over the past few years. A lot of the existing confusion from end users and even some vendors and analysts comes, I believe, from overstatements and misinformation by both SaaS vendors who are zealously trying to drive their change into the marketplace and from traditional SW vendors who are of course trying to protect their existing business. At an event recently I heard an executive from a leading SaaS vendor make the statement that SaaS was replacing all on premise software going forward and that the world would be all SaaS soon…now while this might sound great for his business if not for his credibility, reality is very different, the world of enterprise software is hybrid delivery with both SaaS and on premise (and appliances, on premise subscription, etc.) co-existing for the foreseeable future. I also believe that we may have issues with defining metrics for new business models that are actually only relevant in older models and that those issues are creating some of the misperceptions on both sides of this argument.

So let’s get to the specifics of what I’m referring to here. The argument that I hear from both sides on return on investment (ROI) and total cost of ownership (TCO) are maybe only half true or maybe are trying to compare apples and tomatoes, they’re fundamentally different. The commonly held belief, at least from on premise vendors is that while in the short term ROI numbers on SaaS might look good, the overall TCO of on premise makes SaaS a poor investment. SaaS vendors have focused more on ROI because it is an easy metric to use to show significant value to prospects.

From an ROI perspective SaaS vendors (and also any other delivery model that uses subscription pricing and licensing) has a strong short time to value message. The implementation of a SaaS product over an on premise product should be faster, it’s simple math, I don’t have to, as a project team, install servers, databases, middleware, nor do infrastructure testing, manage infrastructure, etc. so the project to get the software configured and up and running takes less time. If it takes less time to implement then by all rights the customer should see value out of the software sooner (again, simple logic).  Also from a ROI calculation perspective it’s fairly straight forward, lets say you pay $1000 up front for a perpetual license, $300 for implementation and then 20% per year for maintenance there after. For the subscription for the same product maybe you pay $36 per month for the life of the relationship with the vendor plus implementation costs of say $200. In the first year I spend $632 instead of $1300, and am using the product sooner (shorter implementation remember) so my ROI is very high (assuming that the software actually generated some value, otherwise why would you do the project in the first place). SaaS looks like the winner, right?

Now look at the on premise counter argument, TCO. The basic premise for the discussion is that if the vendors stay constant and you take a long enough period for that constant relationship, on premise is a much better buy because the TCO is lower. Just looking at the software costs this point is easy to illustrate. Take the same pricing in the last example, and assume the relationship between the vendor and the customer lasts 10 years, how much did the software costs? The on premise cost $1300 + $200 per year after the first year, or $3100 total. The SaaS product costs $432 per year x 10 years or $4320 + $200 initial implementation = $4520…so the SaaS product seems to have cost 30+% more and thus have a lower TCO. But wait, what about other costs? The SaaS implementation had no license costs for database and middleware and no hardware costs up front. That’s not all though, since they didn’t set up a datacenter to support the new apps the customer saves on resources to manage, update, support, train, etc, all of those system infrastructure components. A true TCO calculation for the same 10 year period then could look very different and I’ve completely ignored integration costs and ongoing costs for both systems…that is if that was the real question here.

As I think about the question, whether SaaS or on premise is “better” or costs less, or returns value sooner, or whatever else, I really need to look at a couple of other thing. The first is a business decision to consume software (all, part, or none) as a service and the second is for each application is the functionality and usability roughly equal between the software products (or maybe which software fits my business needs the best) despite the model. In other words a lot of other issues have to enter into the discussion of SaaS or on premise or appliance or subscription priced on premise. Things like is it better for my business to rely on a vendor to deliver the datacenter (or cloud) services to me rather than invest in them myself? Is the required functionality available in whichever model I prefer? Is there business advantage in one product or one deliver model over the other for my business? Is it better for my business to consume software out of my operating budget in a pay as you go model or is it better to invest capital in the deployment of business systems?

Buying and implementing software to generate some necessary business value is a business decision, not a technology decision. The type of technology needed to meet the business requirements comes after defining the business goals, needs, objectives, etc. Each delivery model has advantages and disadvantages financially, technically, and in the context of your business. It’s like planning a vacation, figure out where you are going and what you are trying to do before you buy a map, book a hotel because it’s cheaper and buy an airline ticket to get a great fare or else you’ll end up with a great deal on a flight and staying at a hotel at the wrong destination…I don’t know about you but I’d rather end up where I planned than end up with a great deal on something I don’t want.