Steve Fisher on Salesforce.com's Platform

I'm extremely passionate about the message in the next two slides because they really go to the two key questions that everybody asks when they're choosing a new technology. The first question is: what is the best technology to meet my immediate need? I've got a pressing business need, I've got users that are screaming for something, I've got a huge immediate market opportunity. And what is the best technology to make me productive and really make my customers successful as fast and as inexpensively as possible? And that's what you're going to hear a lot about for this entire day: why, in our view, cloud computing and platform as a service really is a fundamental shift in terms of productivity.

For example, in the traditional model, what you end up doing is assembling a lot of piece parts and putting it together, designing your networks and your data centers, figuring out your application server and your workflow, and all of that. It's all up to you. You don't really get much help except maybe from an expensive consulting partner. But with platform as a service, with the cloud, instead you can just tap into the cloud. You don't have to deploy the data centers; the data centers are there with full disaster recovery and full redundancy. And given the appropriateness of the platform, you can also get things like a full customization engine to customize the data model, customize the logic, customize the UI, an API for integration, an analytics engine, a sharing model, and a workflow engine. All those capabilities that we required, as Marc said, to build our CRM application, they are all made available to you as a service—a radical new idea, as a service. You can immediately start being productive and delivering your application.

But the next slide is really, I think, more fundamental and one that we're not going to be spending quite as much time on at this event. This event is mostly to help you understand how you can be immediately productive with platform as a service. But the second question, in addition to what's the best tool to meet my immediate need, is what's the long-term impact? What's this going to mean for me a year from now, two years from now, three years from now, a decade from now? Because when you're making a technology choice, and especially when you're making a platform choice, that's a very, very important question.

If you think about Salesforce.com today, Force.com today, 43,000 customers, each of those customers running 4, 5, 6, 10, 20 applications. In the traditional model, that would be hundreds of thousands of separate stacks, databases, and application servers of all different shapes and sizes deployed all over the world. And that is really driving the complexity of the IT industry today. That has been the downside of the great breakthrough of the client-server model: it has resulted in this mass proliferation of separate technologies. And so Marc talked earlier about the massive inefficiency of that. And so you see things like moves towards consolidation and virtualization and all of that to address problems like that. And of course, whenever any one of those applications, you need to scale it up, well, that's a big project because going from 50 users to 5,000 users, typically, that's not going to come for free.

But contrast that with the multi-tenant model. Because in a multi-tenant model, there aren't hundreds of thousands of separate stacks; there is one infrastructure, there's one stack that all of those applications and all of those customers are running on, with the full power of the single-tenant model but without the same underlying technology architecture. It's a fundamentally new multi-tenant architecture, architected in a different way from the traditional model to be able to deliver the power of the single-tenant model, the flexibility of the single-tenant model, for running on shared infrastructure.

So what does that mean? Well, obviously, things like consolidation projects and the need to deal with resource utilization, well, that just goes away because it's already consolidated. Resource utilization is very high; it's a single infrastructure, a shared single infrastructure. Things like, well, if you write an application for five users and it's successful and you want to make it work for 5,000 or 50,000 or 500,000 users, again, it just works because it's riding on that shared infrastructure. Even simple things like deploying development environments, testing environments, staging environments—in a traditional model, that is expensive, it's costly, you cut a lot of corners. In a shared multi-tenant model, you just click a button, and because it's all shared infrastructure, a couple of records go into the database, and all of a sudden you've got your development environment, you've got your testing environment, you've got your staging environment. All those issues go away.

But I want to talk about really what I view as the three fundamental challenges facing technology today: security, performance, and upgrades. This is where we spend massive amounts of time, and this is where multi-tenancy is fundamentally different. So think about security. You can imagine that when we talk to potential customers like Merrill Lynch or Cisco or Citibank or Dell or Symantec, of course, they all take security very seriously, and they all have their security organizations whose whole mission in life is to make me and my team build the platform look bad by putting out all the flaws that they may uncover. And that's great, because when they go in there and they do all their testing and they try and find all these vulnerabilities, occasionally they'll find something—it's technology, occasionally they'll find something. But unlike a single-tenant model where the vendor would find something and figure out a patch and release a patch, and it's up to every single one of those customers to do its own individual testing and decide whether or not to deploy the patch or not, in a multi-tenant environment, the vendor—us—we figure out the problem, we figure out the solution, we deploy it. It's multi-tenant; everybody gets it immediately. So the result is that our multi-tenant environment actually gets more and more secure over time. It's counter-intuitive; we're so not used to that. We're used to having things kind of becoming scattered security, and some places are secure and some places aren't. But in a multi-tenant environment, it gets increasingly secure. Every single time we fix a problem, every single time we close a vulnerability, everybody benefits.

Think about performance. This is where the single-tenant on-premises model is really broken. Because whenever a developer writes some code that they ship out to you, they build in lots of assumptions about how you will use that code. And the more flexible that code is, the more that code is a platform as opposed to a canned application, the more assumptions they build in. And the result is that in order to make it perform, especially over time, that's an effort, that's a project, that's your responsibility to tune that. And it's extremely unlikely that you're going to be able to get the developer that wrote that code to go to your site to tune that application. It's going to be somebody else with less knowledge than that person. But in a multi-tenant on-demand platform as a service, it's very different. The developer that wrote that code is actually able to see how that code is being used. And so in platforms like ours, we are constantly tuning, constantly fixing the performance of that platform. And the result is that, as you saw in the chart that Marc showed, even though the number of transactions grows exponentially, it gets faster and faster and faster. That's because in a multi-tenant environment, it's a learning system. The application, the platform, the people that wrote it are all partners with you and able to constantly tune, constantly improve, constantly make it better, as opposed to having it be broken, having it be disconnected. So more secure over time, faster over time.

But the biggest benefit is around upgrades. So Marc asked earlier how many of you are Salesforce customers. If you could again raise your hand just so that I could see. Lots of people, Salesforce customers. A week ago, we released Summer '08. How many of you that raised your hand had to redo your customizations because of Summer '08? No one. Better. How many of you had to rewrite your integrations because of Summer '08? Nobody. How many of you had to retrain your users because of Summer '08? Nobody. That is amazing. That is so different from think about all the other enterprise software that you have. Customizations break, integrations break, user experience changes. It's a project. It's six months, 12 months, 18 months, hundreds of thousands of dollars, millions of dollars, and you're faced with this horrible choice of, I really want that new feature, but I can't install that software; that's a big effort. In the multi-tenant model, that choice goes away. Seamless upgrades. The upgrade happens, Summer '08, you saw in the room, no impact except you get a bunch of new tools in your toolkit that you choose when to deploy. And the result of that is, because the cost of change goes from expensive to zero, tremendous innovation. We do major releases every four months. I think if you look back at where we were a year ago, where we were two years ago, just unrecognizable in terms of the pace of innovation. And yet every single one of our customers can absorb that innovation and select to use the pieces that make most sense for them.

So I think when you look at the really the core problems that face the technology industry today—things like resource utilization and consolidation, security, performance, upgrades, innovation—these are core issues. And fundamentally, a multi-tenant platform, a multi-tenant architecture, a next-generation architecture is the right long-term bet because it addresses all of those core issues while at the same time giving you a much more productive environment for you to be successful.