Can't get no SaaSifaction

The first session of the Enterprise Software Summit focused in on SaaS, which is a topic probably on the boilerplate for all conferences focused on enterprise software these days.

The group discussion that ensued was really good and touched on a broad array of topics. On one level I am surprised that anyone would even suggest that there is a debate about SaaS but it's entirely apparent that there is a lot of confusion about what SaaS really is.

On one level SaaS is assumed to be straight up "salesforce.com", which is of course a software application that is delivered as a subscription service. However, this is not only a simplistic definition, it is also in accurate to suggest that SaaS is limited to what SFdC is doing.

The enterprise software industry as a whole started a migration to SaaS as embodied by SOA well over 3 years ago. Any debate that is remaining has to do with how SaaS will be realized. On one end of the spectrum are those that argue that all software should be delivered as a service by a third party who provides that service as some kind of subscription licensing model. On the other end are those that say software should remain an on-premise implemented package of "stuff" that is knitted together with SOA principles.

In the middle of this debate are people like myself who want enterprise software delivered in multiple channels and embracing of the principles of Soa as opposed to soA. What I mean by that is the distinction between those who want to build software the way that they always have, except now they want to use all the SOA labeled technologies, and what I think is the endgame, building applications that are highly personalized based on application services that are provided by a multitude of vendors. The former is soA while my preference is Soa.

Whether an application is delivered as a service or as an on-premise installed application is something best left to customer choice. I like hosted models, and would prefer that the industry move forward on multiple flavors of hosted model, whether multi-tenancy or some kind of hybrid that shares a common code base but enables componentry extension or outright replacement. I think this is kind of what SAP is hoping to accomplish with our on demand CRM offering launched last week, but in all honesty I was disappointed in that launch because it sounded too much like us telling customers that when they "grew up" they could graduate to the "grown up table with full blown CRM". This misses the point, but it's better than what we were offering the week before - nada, zip, zero.

To summarize the point that I just made about confusion on SaaS, it's clear that there is a SaaS development track, a software delivery model choice, an economic licensing/support model, and a broad data/process integration issue that underlies all of the above.

The other topic that we spent alot of time talking about this morning was how centralized IT is not the way companies are going to market anymore. I don't think it would surprise anyone to suggest that while centralized big company IT is something that is very necessary, it is also not the place you go for innovative forward looking applications, at least as a user or business manager you don't go there.

What is happening increasingly is that divisions and departments are exercising their ability to deploy packaged applications as a hosted service or an appliance and sidestepping IT. A number of the companies in the room today testified about how they were going to market with a departmental sell and quite honestly I think this trend will not only continue but accelerate as packaged applications become more "packaged" and much cheaper due to commoditization of infrastructure technology. This is probably bad for SAP in the long run because we are forced to fight a battle with multiple fronts, not only competing against competitors but also hardened departmental projects that not only are "there" but also working.

Lastly, while it wasn't proclaimed outright in the session this morning I think a lot of people are quick to think that enterprise software as it has existed is dead. Not only is this conclusion not based in history (hell, there's still a big business in developing AS/400 software and IBM hasn't sold one of those boxes in years) but it also presums that incumbant vendors haven't prepared for the shift in how software is built and delivered. Enterprise software is notoriously difficult to predict, I remember back in 1995 the raging debate was all about who's flavor of client/server was going to win and then Netscape came along, of course it took the better part of 3 years for full blown web-based enterprise software to happen but it did.


Technorati Tags: enterprise+software, SaaS, SOA, on+demand

Comments

[Zoli Erdos]

Jeff, I agree with most of your points, including the two ends of the spectrum, except one bit: On one end of the spectrum are those that argue that all software should be delivered as a service by a third party who provides that service as some kind of subscription licensing model.
Isn't that the ASP model of the late 90's that failed spectacularly? I think if you replace third party with the software vendor, you're much closer to the model that works today.

[Zoli Erdos]

oops, it's little less readable, I lost the html for emphasis.

[jeff]

Zoli,
I think the big difference between the Corio's of the world (which still end up getting sold for a pretty tidy sum) and the salesforce.com's of today is multi-tenancy. This single innovation introduces economics that appear to make the on-demand hosted model economically viable.

[Dennis Howlett]

I don't agree with large parts of this analysis. But on one thing I think we are agreed - multi-mode delivery. This needs to be worked out according to individual requirements rather than at the behest of the software vendor's preferred licensing model. That's a mega challenge for the industry.

[jeff]

Dennis,
Can you tell me what you don't agree with?

[Zoli Erdos]

Jeff,

Back then I was with a SAP-focused SI which experimented with the ASP model (the term did not exist). What made it economically unviable as a third party, we were perceievded to be the all-in-one ASP. In that concept we'd have to install, train people to support a bunch of other systems the customers used along with SAP, since the expectation came all from one source. You can't do that unless you already are scaled up, but you won't scale up without it - catch 22. I think THAT killed independent ASP's and that's why the Software vendor becoming the ASP itself (OK, I know, we call it SaaS now...) is a viable model.

(Now I wonder if I have to coCo this again - my guess is not, since it's already monitoring this thread)

[jeff]

Zoli,
The late 90's ASP's were in an untenable position IMO, not only did they not get any breathing space from vendors but they had to deal with the complexity of applications not designed for that type of implementation.

I suspect you would agree that today's multi-tenant hosted apps are radically less complicated than what you dealt with then.