Oracle president Charles Phillips passed through London recently and spoke to a handful of journalists about the company's prospects and direction, including its SaaS strategy. I noticed Angela Eager's write-up for CBR a couple of days before going off on a week's vacation, and although it had me mildly fuming I couldn't find time just then to blog about it.
Last week, Stuart Launchlan's report of the briefing appeared on AccountingWeb, sparking a discussion among the Enterprise Irregulars that greeted me when I opened up my inbox this morning (for a flavor of what transpired, see Dennis Howlett's comment at the foot of Launchlan's article, which begins, "Wearing my curmudgeonly hat inside the Enterprise Irregulars Google Group, I described Charles' claims as outrageous that should be committed to what I call the Orafice Hall of Shame ...").
Now it's my turn to pick up the curmudgeon's baton. Phillips' remarks are full of ... well, what can I say? — the most egregious misconceptions (or should that be misrepresentations?) of the SaaS model. Sadly this is typical of the kind of rubbish that gets trotted out whenever an unenlightened conventional software vendor decides to assert its familiarity with the SaaS model (SAP and Microsoft Business Solutions are culprits too).
I take issue in particular with four of Phillips' assertions (all quotes are from the CBR article):
#1: Customers want to replicate data
Phillips believes the very nature of SaaS will drive demand for on-premise databases. "SaaS is very database intensive. Normally people do not want all their data resident on an on-demand product. So if Salesforce.com is hosting data for a large company, they are forcing them to create a replicated database behind the firewall, which means that companies are creating more and more databases," he said.
If there's a scale for measuring egregiousness, this self-serving nonsense certainly shoots off the high end. Is Phillips saying the SaaS model suddenly reminds customers of the hitherto ignored need to keep backups? Or that it leads them suddenly to discover that they cannot analyze data across multiple application silos without keeping a separate copy of some of the data (strange that a man in his position should not be aware of the concept of data warehouses, but we'll let that pass). Or perhaps he's attempting to imply that organizations that decide to have an external provider hosting their data don't really mean it, and are gaily budgeting for another server farm's worth of RDBMS licenses just to be on the safe side. Puh-leese.
#2: Multitenancy doesn't matter (people want separate databases)
As for the issue of multi tenancy, which is one of the core tenets of SaaS, he suggests it is irrelevant from the customer perspective. "None of us do [SaaS] in exactly the same way, we use different architectures. Multi-tenancy has nothing to do with on-demand. [It is] a convenience for the vendor. Whether they put all customer data onto one database or onto multiple databases is of no value to the customer. In the enterprise it is the opposite. They do not want to put all their data [into the same database] as their competitors. It is illegal in some industries," said Phillips.
A lot of conventional software vendors try and persuade themselves that multi-tenancy doesn't matter. And Phillips is correct when he states that it's a convenience for the vendor. In fact, I would go further and say that it's a liberation for the vendor. But for that very reason it's of immense value to the customer, because it significantly reduces the cost of delivering the application (and therefore the price). Even more importantly for conventional ISV developer teams, it prevents them slipping back into customizing code for individual customers, as Siebel's CRM On Demand division has done, thus hobbling its competitiveness to the extent that it is just ebbing away as a market player.
Note that multi-tenancy doesn't prevent vendors from hosting customers on their own individual servers. Read NetSuite's S1 filing and it makes clear that smaller customers are hosted on multi-tenant servers while larger customers are hosted across multiple servers dedicated to that individual customer. Many other on-demand SaaS players adopt this model, including RightNow and Workday. The point of adopting multi-tenancy is not that it forces SaaS vendors to host all their customers on shared servers, but that it gives them complete choice as to how and where they host their customers. The majority of customers, of course, really couldn't care less where their data is kept so long as the vendor makes sure that it's secure and available. Those for whom it does matter can easily be accommodated within the multi-tenant model.
#3: Customers want to own software
He maintains that customers make two decisions when it comes to software, one concerns ownership, the other the delivery mechanism. "We have customers who want to own the software, pay for the license and then decide if they wanted to run the software themselves or have us do it for them. They really want those decisions separated. They want to be able to switch, bring it in-house or mix and match. It is multiple choice. The issue is whether they want the running of it [to be] outsourced or not."
I've had run-ins before with proponents of this fallacy that customers want to 'own' software. What on earth is beneficial about owning a licence (not even the source code) to software that forces your business processes to conform to every other licencee of the same software? (unless you customize it, in which case upgrading will cost you an arm and a leg). There is no conceivable benefit to customers in 'owning' software on such terms; all the benefits accrue to the vendor — including being able to tell customers that, having bought it, it's their responsibility whether it works or not. Perhaps the most breathtaking achievement of the software industry is to have persuaded customers that it could possibly be in their interests to acquiesce to this state of affairs. What customers really want to own is their business processes, and the SaaS model gives most customers much more effective day-to-day control than any conventional software package.
#4: SaaS is just a hosting option
"We can give customers a choice in that we offer and integrate both on-demand and on-premise and make it look like a single architecture. From a customer point of view they can have CRM on-demand, or on-premise, or both. We can sell to Siebel customers who want to do some things on-demand and keep some things in-house. Whatever decision they want to take, we can accommodate it," he explained.
It's an established tenet that turkeys don't vote for Christmas. By the same token, conventional software vendors maintain that SaaS is just a different way of delivering the same software. To admit anything else would be to own up to their own obsolescence. Understandably, this is something Phillips resolutely refuses to do.