Is SaaS Right for You?
For some users, the cost (free) might trump availability issues. For others, though, not so much.
Like most editors (and developers I know), I'm a control freak. And little brings out my inner control freak like the notion of Software as a Service (SaaS). Being unable to access material, except when in a connected state, is a daunting limitation to accept. I have a range of misgivings about such apps, apart from potential availability issues related to the Internet. What about security? How willing am I to trust the service involved with data? Is Google innately more secure because of the investment behind it? Or does that just make it a bigger target? Or, what if the company goes belly up tomorrow? Do I simply lose access to the data that might have been hosted?
But even I accept there are cases where this model makes sense, and I've been finding areas where I'm willing to cede this control. For example, online photo hosting, especially as practiced by Google and its Picassa service -- seems like a great deal to me.
The price -- free -- is hard to top. Some of limitations of Picassa don't bother me much because they're easy to work around. Yes, I can only post new photos when I have a live Internet connection and the service is up, but posting something in a timely manner is different when it's photos of the family, versus, say, an edit for the magazine, which needs to be to my managing editor 10 minutes ago. Do service-based sites go down? Of course they do. Everything breaks down at some point, but I'm not counting on three-nine availability for my online photo album anyway. If it's down when I go to use it, I'll log in later. If it's down when someone goes to look at pictures I've shared, the person who tries to look at them will just log in later.
One thing that makes Picassa work for me is that it's a non-mission-critical application. If Google stops offering the service tomorrow, I still have my pictures, stored locally. I can choose to repost them to a different service, or to share them through some other mechanism.
I feel differently about a business-productivity suite. I find it convenient to be able to compose documents when I'm offline. The wireless connection in my home hasn't always been reliable, and not being able to work when my connection does go down might impact my ability to do my job. Now take that scenario and apply it to an office setting. If your network goes out, and users rely on service-based productivity suites, do you lose more productivity in that downtime than you gain in potential savings by relying on a service-based solution?
An office productivity suite is one thing -- a big thing, even. But take it to the next level: Would you trust your company's data to a service? My initial reaction, edited to excise profane language: "Hell no!" Yet Amazon (SimpleDB), Google (Google App Engine), and Microsoft (SQL Server Data Services [SSDS]) all have released recent betas for off-premises data services. Has the world lost its collective mind? Well, maybe not. Roger Jennings makes the business case for such services in his overview on Microsoft's SSDS in this month's issue:
"Accommodating rapid user growth and fluctuating traffic patterns ordinarily requires drastically overbuilding Web infrastructure to reduce latency when traffic spikes occur.
Renting scalable resources in existing data centers with a monthly payment based on gigabytes of data stored and up or downloaded is far preferable to making large up-front capital investments, especially for startups."
The negatives and concerns about this kind of business model still exist -- and they are many -- but the financial justification for assuming such tradeoffs begins to make sense to me in this context. You might not know how robust or large the infrastructure to support your new widget store must be, and making the wrong forecast, in either direction, could be financially crippling. The SaaS model helps you allay some of that risk, and I could see that same argument working in many other scenarios, as well. (For more on SSDS, see Roger Jennings' overview of the subject).
Talk Back: Would you trust your company's data to a SaaS implementation? Tell me at firstname.lastname@example.org.
Patrick Meader is editor in chief of Visual Studio Magazine.