Sunday, April 06, 2008

Platform as a Service Enables Cloud-Based Development While Accelerating Role of SaaS

Transcript of BriefingsDirect podcast on platform as a service and on-demand applications deployment trends.

Listen to the podcast here. Sponsor: Bungee Labs.

Dana Gardner: Hi, this is Dana Gardner, principal analyst at Interarbor Solutions, and you’re listening to BriefingsDirect.

Today, a sponsored podcast discussion about platform-as-a-service (PaaS). We're looking at an entire lifecycle approach to services on the Web for development, integration, and deployment. We are going to be talking about Bungee Labs, the sponsor of our podcast and take a deep look at how they approach PaaS.

I think it’s important for us to get into this topic to understand its wide-reaching implications. PaaS is really taking the best of some of the old and the new that’s now available to developers and architects -- and it really puts a new spin on Web-oriented architecture (WOA).

For example, many developers can use tools that are fleet and easy as they approach rapid application development (RAD) and Agile development. But when it comes to the flip side -- the deployment -- there are still gaps in many cases a hand-off. There is also an opportunity, now that users are increasingly on the Web, to use mashups and open APIs.

To take advantage of data independence -- data that can be acquired, used, and mashed-up from the variety of sources -- including services-orientation from inside the firewall, from within legacy applications. What's nice about the PaaS approach is that the scale on the deployment can go up very rapidly and scale down efficiently in terms of cost.

In many respects, we are now leveraging the incredible value that comes from not having to pay upfront in capital expenses for applications and infrastructure, but instead can take advantage of the pay-as-you-go, metered approach.

To help us better understand PaaS, we’re joined by Phil Wainewright. He's an independent analyst, the director of Procullux Ventures, and a ZDNet software-as-a-service (SaaS) blogger. Welcome to the show, Phil.

Phil Wainewright: Good to be here again, Dana.

Gardner: We are also joined by Alex Barnett, the vice president of community at Bungee Labs. Welcome to the show, Alex.

Alex Barnett: Many thanks, Dana, for having me on.

Gardner: Now, as I mentioned, this on-demand platform value brings together a lot of the old and the new. But in some regards, the current state of development for applications and Web services is still a tricky business. We're still in a period where friction between the developer and what the operational environment is expected to be.

There are often integration issues. We are also having to deal, of course, with on-premises software acquisition and downloads, licenses, upgrade paths -- just maintaining the software on premises for development and for testing.

Phil, could you help us understand what is it about the current state of development that will benefit from moving all of this into the cloud?

Wainewright: One of the factors is that IT is getting more and more complex, and obviously there is a lot of emphasis on governance and oversight and so on. So, provisioning a new system for a development project can be a very long-winded process.

On top of that the world seems to be moving fast ... because of the Web, because it’s connecting us in much more immediate ways. And therefore, business people want to have much more record intervention, and they need automation of new processes a lot faster. You’ve got these two processes: On one hand, it’s getting more and more complex, difficult, and time-consuming to bring new projects online. At the same time, there is this pressure to have those new projects faster and faster, and to have an adaptable and agile development environment. It’s a collision course, really.

Gardner: And I suppose with the way teams are dispersed now, and with people leveraging globalization and offshore development, that you want to bring groups of people together without also having to maintain and oversee the on-premises software at each location that’s supports new development along with application lifecycle management (ALM) ?

Wainewright: Yes, that’s absolutely true. The web enables this collaboration, whether you are developing on on-premises platforms or in the cloud. You want to do it via a distributed team, because that’s the way that you get to this scale, and you can get the best minds on a project. And so you are going to be doing that collaboration in a kind of Web-connected environment anyway.

Gardner: I suppose it’s also important when we look at the speed of development to appreciate that three or four folks in a garage in Northern California can come up with a mashed-up service and go out and create a social networking company, for example, quick and easy -- and at low expense. And then, here you are in an enterprise, taking six months to work through a requirements process. It seems as if something has got to change.

Wainewright: This is a rewind back to the late '90s when people looked at the Web and they thought, "Well, how is it that my daughter or son can look up the information of some arcane homework project in a couple of seconds, while it takes me three weeks to find out what's happening in my own business?"

Now what we are seeing is my daughter is able to hook out to these applications from some guys working off in a garage for Facebook, and do it all in a couple of days. And, again, I am kind of trying to get similar functionality on my corporate network -- but it could be 18 months before I see it.

People are asking, "Well, why, why is that? Why are we so far behind what these people in garages, these teenagers, are able to do so much out there on the Web?"

Gardner: That’s a reason why we are seeing very rapid uptake in what's known as Enterprise 2.0, WOA, mashups, rich Internet applications (RIAs) -- and it's not just startups, but really starts to move across the board. It’s just a faster, better way of getting a fairly large class of applications going -- and also managing the integration of data on the back-end.

Wainewright: Well, let’s be cautious, Dana, because I think that a lot of people are experimenting with it for the moment. I wouldn’t say enterprises are necessarily doing things that are business-critical or mission-critical right now. They are still working out what the utility is, and what the risks are from this model.

Gardner: Let’s take that over to Alex. Alex, tell us a little bit about how you perceive the rates of adoption around SaaS and WOA. Do you see this as something that is just building; and what about that issue about risks security, control and management?

Barnett: The context from the big picture point of view is that everything we know is moving to the Web. And what that means in tangible terms is that businesses and service providers and software companies are providing layers of functionality and data that are native to the Web or are Web-oriented.

Things like Web services and even less-sophisticated and increasingly popular things using REST and XML offer interfaces into the functionality and data. This ongoing trend is being driven by a frustration in what we’ve inherited from the previous generation of computing, where everything was essentially behind a proprietary protocol or proprietary set of APIs. This required that you to buy a platform set, or toolset, across the board, just so you were able to get the data that you owned as a company.

If you think about customer relationship management (CRM) systems or a certain kind of database -- they are in silos, they are disconnected, or they are very expensive and require lots of proprietary knowledge in order to be able to access the data, and therefore the value.

What we are seeing with things like WOA is a materialization of how we get out of that frustration as an industry -- how we get out of that frustration from business. We want the data and the business intelligence from it, and to be able to get at that from a business perspective.

The IT managers feel the pressure to be able to more rapidly to develop applications and access the data that are based on and being made available through Web APIs. And developers are able to then connect, develop, and build-out new applications based on distributed data, on distributed functionality and react to the business needs.

If you look at, for example, at the end of last year, you would have seen that there were something like 500 Web APIs catalogued. There is a combination of commercial and public APIs. And now, here we are in April and we’ve got 650 APIs from the public and commercial space. So exactly the same trend is occurring behind the firewall, within organizations, as they create these layers of connectivity and programmable end-points to gain functionality and data.

Wainewright: It is important to say that the interfaces being published on are not just some stupid little kind of stock market data feeds or what's the weather in my area tomorrow, the sort of thing we saw in the early days of Web services. These are actual, complete and functional API sets of the kind that an enterprise calls to applications like, and to many other resources.

Gardner: So we have some significant new opportunities for taking advantage of the Web with these mashups and the data portability. And, again, we want to take advantage of the old as well as the new, with the old being characterized by lower-risk control and management.

So what does PaaS do conceptually, Alex, and how does that bring together the best of the old and the best of the new?

Barnett: If we define "old" as being that you are locked into a stack, that you aren’t able to get up through using open standard protocols such as SOAP, or even just standard XML interfaces, then that "old" is typical to get around. That’s what we’re seeing in terms of investments to upgrade system to unlock that value.

Gardner: I guess by "old" I mean enterprise-ready and mission-critical, that’s what I mean.

Barnett: Right, and once they’ve got to the point whether they have allowed that services layer to occur, the question is, "How do we get to the next step? How do we quickly derive value from the systems we have?" What we hear businesses crying out for is getting at the data, getting at the applications that are open.

Now, there are some companies that are moving ahead very quickly with this, being able to take the leap of faith of providing CRM-type data on a hosted service. Or, they may want to maintain it behind the firewall, behind their existing systems, and then just be able to provide limited, yet secure, access to that data through applications that are inherently secure in terms of their architecture.

Wainewright: What we're seeing develop at the moment is kind of a two-tier information technology. On the one hand, you have all of the existing on-premise, legacy applications and all that data that Alex was describing. It's locked away, and IT managers are really puzzling over and grappling with the issue of how to unlock that data. How do they make it more accessible? How do they build more agility into applications infrastructure?

They are looking at things like service-oriented architecture (SOA) and other ways of connecting and integrating the data, while automating business processes within the organization. So that’s one tier.

The other tier that’s developing is this Web-oriented tier, all of these APIs and in-the-cloud resources and applications that are out there on the Web. To take advantage of those connections, you need to build a completely new infrastructure, which is different from the existing infrastructure within the firewall. It has to cope with connecting to external resources, and it has to have different kinds of security, different kinds of identity management.

Building a robust infrastructure to do this is very, very hard. That’s one of the reasons why a lot of enterprises are holding back, and are therefore missing agile opportunities. That’s one of the roles that the PaaS can provide.

Barnett: What they have tried is failing, or they haven’t got the right level of investments, or they have other priorities.

Wainewright: Yes. And, we see some tremendous disasters happening where people in e-commerce are, in a very limited way, opening up to customers to transact with them. And they have to do that. They are losing customer data, because they're not getting the security right. So, yes, people have to be wary of that whole area, that Web tier -- because it’s full of pitfalls and traps.

Gardner: Okay, we’ve established that the tide is turning to the Internet, that there are some great Web-based services available, that technologies are now bubbling up to allow for better and easier connectivity. And yet, there is still a need for the right platform and the right infrastructure to make this all mission-critical and enterprise-ready.

So let’s get into PaaS as a possible stepping stone that, in a sense, bridges the best of the Web-oriented architecture and the available SaaS and the APIs-world with what developers inside organizations -- be they ISVs, service providers, or enterprises -- need to make these approaches acceptable and within the acceptable risk parameters.

I noticed that Bungee Labs does not call this "Development-as-a-Service" or "Deployment-as-a-Service" or "Integration-as-a-Service" -- but "Platform" as a service. Alex, give us the primer. What does "Platform-as-a-Service" really mean?

Barnett: That’s what we are trying to define at Bungee Labs. PaaS is one of those terms that we’re going to be hearing more and more. And they are going to be different -- varying levels of definition and interpretation of what that means.

But what we’ve done is put a stake in the ground in this respect, and then saying that in order to really be a PaaS -- and not just any one of those single pieces that you’ve mentioned plus more individual pieces -- that you need to be able to provide the end-to-end services to really call it a "platform."

From the developer’s standpoint, which is the development cycle, this means the tools that they need to develop applications, to be able to then test those applications, to be able to connect to Web services and to combine them, and to have all those kinds of capabilities -- and to then deploy and to make those applications instantly available to the business users.

Literally, we mean a URL that is the end-point for the end-user. From that, they can start consuming the application.

So, PaaS means having an environment in which you deploy inherently and have built-in scalability, reliability, and security. Once you’ve deployed your application, you know that you don't have to take care of all the infrastructure in the datacenter and the capital investments and the bodies that are required to make it scale when newer applications increases in use.

There is also the ability to connect to the various distributed data sources or functionality that the application needs to be able to consume. You can get that inside of that platform, the ability to be able to do that in a Web-native way, and so take advantage of the architectures we descried earlier, such as SOA.

There is also the ability -- and we touched on it earlier -- for developers to be able to collaborate on projects that are built-out in the cloud. They can share code, check in code, do all the standard revisions and collaborative-type functionality that developers need when they’re working on projects with teams distributed across the world or across your offices. And they can do this without having that entire infrastructure on-premise.

And then, the last, but critical, piece is having deep instrumentation and an analytics ability around the use of the application -- of how it’s being used, of where the connections are -- right across the board from the "glass of the window," the browser, for example, and right on through to the Web services in the CPU, or the rest of it.

As a result, you are able to understand performance. You are able to understand your billing, if it’s a billing proposition that you have. And all of what I described is comprised within six pillars [of Bungee's offerings]. All of that is delivered and available purely as a service, so there are no on-premises requirements for any of those components across the development and platform used in a utility model. You use it as much as you pay for, or as much as you use in a utility-based model -- all in the cloud. No bit needs to be installed on any machine at the enterprise in order to take advantage of all those Web services and functionalities.

Gardner: For our listeners who are just getting used to this concept of PaaS, let’s just get right in quickly and describe what Bungee Labs is. It’s a young, innovative company. And you’ve come out with a service called Bungee Connect. This is essentially one place online where you can go to develop, mash up, and access data, to put together Web-based applications and services, and then instantly -- with a click of a button, and perhaps I am oversimplifying -- develop and deploy in basically an integrated continuum. Is that correct?

Barnett: Yes, and provide very rich user experiences as part of that, with highly interactive application functionality. We’ve built out essentially that stack that I’ve described earlier. We've made that available for organizations to take advantage of. We're specifically targeted at developers who really want to be able to build very sophisticated Web applications that leverage orchestration workflow around connecting to Web services.

We are not in the business of being able to provide non-programmers with the ability to do these nice simple mashups.

Gardner: Well, if you can do that, let me know, because that would be a very good trick. I am sure the world would love to have development by anybody!

Barnett: Yeah, and that’s a great dream to be able to have, but inherent in that is inflexibility, because you are simplifying it all for the end-user. What we really offer is for the developers who are tasked with building sophisticated Web applications to do just that, deploy that, and then deliver very rich user experiences out on the Web.

Gardner: And to be clear, this is not just open source. This is commercial code, if they wish. The people who develop on this system, that code is their intellectual property. Is that right?

Barnett: The intellectual property of the code that is developed by the developers is absolutely their own intellectual property and remains so. We do have a community side of things that allows developers -- just as in the open source world -- to be able to share code and even entire applications as open source running on our grid.

But in terms of a company, it’s entirely their intellectual property that they developed and they are able to literally export the code. And if they want then re-factor that for a different kind of a grid or runtime, it’s their property.

Gardner: Phil, how do you see the relationship between PaaS and what Bungee Connect is doing, and then the larger SaaS trend? Do you see a relationship of one aiding and abetting the other? Or are they in separate orbits? How does that work out?

Wainewright: I think they are very much in a similar orbit. And to an extent, I don't think of PaaS as being part of SaaS or vice versa. It’s just everything moving to the cloud. These are two examples of that happening.

One of the things I want to highlight, as Alex was saying, is the useful experience. When people start developing for the Web, for the cloud, then it’s not just building the infrastructure -- it’s also learning what is involved in writing applications for that environment.

There is much more emphasis on the user experience. There is much more emphasis on reusing what other people have done, whether it’s by mash-ups or by reusing other people’s code, as opposed to reinventing the wheel every time. There is much more emphasis on developing applications and programs that can adapt and change to future opportunities in business conditions.

All of those things also have to be learned, at the same time as building the infrastructure. Using PaaS enables you to tap into that shared expertise in a way that you can’t do, if you try all by yourself.

The other thing that’s happening here is that we’re connecting into the resources of the Web, and getting onto the Web, so that we can interact with partners and customers and connect into those other Web resources. This is what we're really expected to do as businesses today, in order to stay competitive. So, there’s a tremendous pressure building to be able to do this kind of thing.

Now, there are three ways you can get onto the cloud. You can go to a cloud-computing provider and basically build your stuff in that cloud, which gets to some of the infrastructure, but, there's still the issue of how do I write applications in this environment and connect to other client resources.

Second, you can go to pure SaaS whereby you get a ready made application and you can do some customization, but there are going to be quite a few gaps around what that provides and what you actually want to do. There are going to be quite big gaps in terms of integrating that into your existing on-premises applications and to the other client application that you use.

Third, where PaaS comes in, it allows for the ability:

A) To get much faster to the custom applications that you need to build for that environment

B) To do the integrations to fill in the gaps and to access other SaaS applications and services, and to patch and connect back to the existing on-premises applications.

Gardner: Well, great. Now to the point a little earlier that if you read a lot of the blogs that are out there you might think that this is all widespread. But PaaS is just in its early stages. Yet one place where SaaS has really become quite popular and in a full enterprise usage is the CRM space.

First, why is it do you think that SaaS has taken off with CRM, and then secondarily, what is it about the nature of the data in CRM that might be something that this PaaS approach might be well suited for?

Barnett: In the early days, CRM took off because SaaS automation allowed a sales manager to get to the functionality he wanted. In the eyes of the IT manager, he said. "Well, okay, this is just a standalone application inside of The sales department is not going to hurt anyone else, so go ahead with it." And the sales manger just signed up on the credit card, and perhaps didn’t even tell IT about it and got on with it.

So it was very easy to establish on-demand CRM. Now what has happened more recently is that enterprises have realized that they have a lot of on-demand CRM being used. And perhaps they’ve decided that they want to harness it because it does kind of enable them to give functionality to the sales team faster than they can, if they built it themselves.

But they would have to do that in the context of their IT infrastructure, and they are looking at things like integration and to make sure a customer record in the on-demand CRM system is the same as the customer records in the ERP system. If a salesman closes a prospect, then how do you translate that closing from the CRM system into an order on the ERP system of records so it gets invoiced and the salesman is properly compensated? How do you then take back the data and functions into the SaaS system?

Gardner: So, we have a lot of fast-changing data that is actually essential to a business. This is what makes their sales happen and how they then fulfill those sales and get them into the processes that their back-end systems will perhaps manage and drive to create the actual products and services. This is certainly mission-critical. It’s distributed across a number of people, with many of them scattered and mobile, and it’s all quite dynamic.

Barnett: Right, and they need to do even more functionally. The on-demand CRM providers like NetSuite,, Oracle, and now we’re hearing Microsoft Dynamics, are providing Web services layers over the functionality that they provide to the end users.

We’ve always been able to do a certain level of functional customization around those applications, but when you have the Web services that provide access to the data -- on the programmatic level -- you gain a whole new opportunity to merge, in terms of levels of customization against an existing CRM application, or ERP systems.

This pushes out the expansibility and the increased functionality of the investments that they’re making. It allows for those services to be able to expand that further in a cloud-orientated, Web-orientated way.

Gardner: Now, back to Phil. What is it about the CRM and PaaS from your perspective that demonstrates a larger opportunity in the market?

That is to say, if we can take advantage of the mashups and services and high level of on-demand CRM -- we can bring PaaS in to help integrate data or the take that data and some of the interfaces and views from a CRM activity and bring it and relate it to either a channel or a supply chain and/or backend systems.

Does that mean that CRM now highlights what will be repeated across other types of business applications?

Wainewright: Yes. I think it’s great for the on-demand CRM vendors, because it really starts to hammer home the benefits of being an on-demand application. Now you’ve got this Web context that you can take advantage of.

As you look to the huge advantages of being able to consolidate community insights into a better application, of being able to connect into API resources and aggregate data for doing composite processes, then that means these are mashups, data mashups and user interface mashups -- it’s all the same kind of thing.

So you gain this ability, and you really destroy the misconception that if you go to SaaS you can’t do customization and you can’t do integration. It means that we’re actually doing better customization and better integration than you are capable of doing with many on-premises systems, because it’s actually a more flexible customization. It’s more cost-effective integration because it’s a shared service.

What previously seemed like disadvantages of the SaaS model can be turned on their head and turned into advantages.

It shows the way to using these applications in other areas. We talked about CRM -- because that’s a very big area that gets a lot of media attention -- but there are other areas that are equally successful with on-demand, such as people management, human capital management, the whole e-commerce area, and, of course, content management. There are lots of opportunities to take this model further.

Gardner: It’s been interesting for me as I look into Bungee Labs and Bungee Connect because it appears that the PaaS value forms a stepping stone to allow more use and exploration of the SaaS applications and services that are available. And the more you use those services -- in a virtuous adoption cycle -- the more you want to customize and integrate. And so you might then look to PaaS as a means for doing that. I think they play off of each other quite well.

Wainewright: They do. And I think probably the biggest takeaway for anyone listening to this is that as a business, you need to know how to work in this Web environment. Your customers expect you to be connected to allow them to participate. Your partners expect it too. You’ve got to be open to the Web. You need to play in this environment.

And secondly, if you are a developer, you need to learn how to do all this stuff, because this is going to be a big, expanding field where the skills are going to be in demand.

A lot of the focus at the moment in IT is on getting all of the internal systems operating together. I think there’s very little budget available for doing a lot of new stuff. So, you can’t afford to build the infrastructure for all of this Web stuff. You really need to go to outside providers to be able to start playing with it quickly, but if you don’t start playing with it now, you are going to left behind.

Gardner: Great points. Also, if you build services and applications on a PaaS approach, they are going to be public facing. Putting it onto a grid, or utility/cloud-type of deployment allows you to then scale up very rapidly without you having to worry about that forklift upgrade of your blade servers or your datacenter. And, of course, you can also scale down if needed if the services impact just a handful of people in a supply chain environment, for example. You can build a service that might be for a small group of people, but at a price-point that makes that worthwhile.

Barnett: I’d say that’s the beauty of what the whole SaaS trend has allowed us to be able to do.

Going back to the CRM example, the sales managers have been able to just start up a test account using their credit card. And, all of a sudden they can start seeing the value instantly of having this CRM on-demand, through a browser. And, then they can just slowly, slowly increase the activity that’s going on there because they see the value and it’s becoming cost effective.

But if you take the same kind of concept and now bring it back to the IT department, they can just try a simple application that has business value, as long as they’re happy on the security side of things. Then maybe they try a composite of two Web services, and provide instant value back to the business. There it is. It’s a URL. It’s secure. It’s locked down, and it’s all the rest of it. And nobody can get access to it except for the end users that you've decided on.

It just slowly builds up. There's no long-term cycle around this massive kind of evaluation, cost and ROI studies, TCOs, and infrastructure investments and projections. You can just like try it out slowly, and if you like what you see and you get in that kind of positive feedback as a developer or as an IT manager or from an end user -- you just slowly build it up and only pay for what you use. And you have the instant ability to scale if it really takes off.

That’s great thing for anybody in business: To be able to try before you buy, without any kind of commitment or contractual commitment or cost-commitment upfront.

Gardner: I am afraid we’re going to have to leave it there, we’re out of time.

We’ve been discussing PaaS and how Bungee Labs has been bringing that concept to market with a service called Bungee Connect, which fulfills much of the underlying functionality for PaaS.

To help us better understand the market opportunity and the trends that support this, we’ve been talking with Phil Wainewright, an independent analyst, director of Procullux Ventures, and a ZDNet SaaS blogger. Good to have you on the show again, Phil.

Wainewright: Pleasure to be here.

Gardner: We’ve also been joined by Alex Barnett, the vice president of community at Bungee Labs. Thank you, Alex.

Barnett: Yes, many thanks for having me on, Dana.

Gardner: This is Dana Gardner, principal analyst at Interarbor Solutions. You’ve been listening to a sponsored BriefingsDirect podcast. Thanks, and come back and join us next time.

Listen to the podcast here. Sponsor: Bungee Labs.

Transcript of BriefingsDirect podcast on platform as a service and on-demand applications deployment trends. Copyright Interarbor Solutions, LLC, 2005-2008. All rights reserved.