I ran into a VC friend at breakfast this morning and we fell to talking about enterprise software. I noted to him that most of the support issues I see relate to configuration, not bugs, and how deploying software in a Software-as-a-Service model would largely eradicate this issue, leaving developers free to focus on development, not supporting deployments.
His counter was interesting:
SaaS is great if the only color you want is black. SaaS doesn't allow customers to fine-tune their implementation. If the vendor decides to update the UI (and force new training costs on the customer), that's too bad. Customers worry about security in a multitenant environment and are screwed if the system goes down, as Amazon's S3 recently did.
SaaS is generally billed as the end of software and the beginning of nirvana. But is it nirvana for customers or vendors?
For the vendor, the benefits are obvious. Build it once, run it everywhere and charge customers monthly use fees. Maintenance is easier, billing is easier, everything is easier. One system to provide for them all.
For the customer, there are obvious benefits, too. SaaS offerings have tended to be lower cost and easier to maintain because, well, there is no maintenance. You just plug in and go.
But this very ease may be SaaS' biggest problem. Even with open data policies, once you leave the service, you leave it. Or, rather, it leaves you. You don't have the software anymore so your data is largely worthless unless you can find another service to import it. You're beholden to the whims of the vendor even more so than in an on-premise software deployment world. Once you jack in it's very hard to get out.
So, it's a trade off, as are all decisions. SaaS offers clear advantages for vendors and customers, but it also presents hefty disadvantages. Whether they're worth it is a decision for the prospective customer to make with eyes wide open.
It seems that perhaps the ideal model for customers would be a SaaS/open source model that provides the software as a service but "escrows" the software for the customer, too, such that they get the source code should they ever wish to leave the vendor. Would the vendor be giving away the farm? No. Customers who tap into SaaS are still getting benefits from SaaS even if they have the source code to run the software on premise: lower costs, ease of administration, etc.
The only difference is that the vendor is giving them a clear exit path. This may not be appealing to the vendor but if not, they should be asking serious questions of their value proposition if it can't withstand a bout of customer choice.