Alex Barnett blog : SOA, Mashup, trends

Time to Define "Platform as a Service" (or PaaS)

alexbarnett — Tue, 19 Feb 2008 09:27:00 GMT

Before joining Bungee Labs last year, I knew they were on to something big. I mean, really big.

A big idea, an ambitious vision: to provide developers with end-to-end development, testing, deployment and hosting of sophisticated web applications as a service delivered purely in the cloud.

Since we announced our private beta back in May 2007, we've had over 1,500 developers sign up. In January alone we had over 400 developers kicking the tires - not just signing up and disappearing, but 400 returning developers, learning, building and deploying out increasingly sophisticated apps on a fast evolving developer platform, requiring no install of anything on their machine - all through the browser.

And since May 2007, the trend to delivering software as a service (SaaS) has been moving at terrific pace. New web APIs are being made available every month and new announcements by start-ups as well established big players are reinforcing and fueling the acceleration to the inevitable world of cloud computing.

As we announce our move from private to public beta today, we've also tried to articulate the new category of product and service we believe Bungee Connect is at the forefront of defining, the category of Platform as a Service, or PaaS, and our big bet is that PaaS is the next big thing.

So what is a "Platform as a Service"?

In September 2006, Marc Andreessen posted his thought provoking "The three kinds of platforms you meet on the Internet" and it got a fair level attention from the web industry. And we took note. We thought what Marc was describing in his Level 3 definition where:

"A Level 3 platform's apps run inside the platform itself -- the platform provides the "runtime environment" within which the app's code runs.",

...was right, but only partly right. Given Bungee Labs' ambition and vision, we felt there was a lot more to Marc's definition of the highest level definition of an "internet platform", a definition more holistic and comprehensive than a runtime.

But we kept focused, kept working on what we were hearing our developers telling us we needed to fix and improve on Bungee Connect, to give what developers are telling us what they really want - a Platform as a Service - to provide everything required in the lifecycle for the development through hosting of full-on, sophisticated and highly interactive web apps, not just widgets.

As we were readying for our next phase -our public beta - we thought it would be a good time to put a stake in the ground and actually define what we mean when we use the term Platform-as-a-service, and thereby describe the comprehensiveness what Bungee Connect has to offer.

So early this morning, our CTO and Founder of Bungee Labs, Dave Mitchell posted a definition describing PaaS in concrete terms.

What follows is a summary of Dave's post, with a selection of my favorite "soundbites" and ideas, but I suggest you read the whole post for yourself - there's a fair amount to consider:

1) Develop, Test, Deploy, Host and Maintain on the Same Integrated Environment.

"It’s time to stop developing “here” and running “there”. Today, most applications are coded in one environment (usually custom-built for that project by a developer), then tested in another, and redeployed to yet another for production...In a completely-realized PaaS, the entire software lifecycle is supported on the same computing environment, dramatically reducing costs of development and maintenance, time-to-market and project risk. A PaaS should let developers spend their time creating great software, rather than building environments and wrestling with configurations just to make their applications run — let alone testing, tuning and debugging them...Also, an end-to-end PaaS should provide a high productivity Integrated Development Environment (IDE) running on the actual target delivery platform, so that debugging and test scenarios run in the same environment as production deployment.

2) User Experience Without Compromise

"A Platform-as-a-Service must deliver compelling user experiences, with all the richness and live interactivity that consumers have been conditioned to expect....Hiccups like software downloads or plug-in installations, browser dependencies and inconsistencies, or local executables break the web model, and are inherently less secure, less maintainable and less user-friendly. In order to be relevant and popular, PaaS must deliver the best user experience available on the web, comparable to or better than conventional approaches.

3) Built-in Scalability, Reliability, and Security

"Developers should be free to build applications with the comfort that the security of customer data, network traffic, source code (intellectual property) and even server hardware is maintained automatically by the platform through-out application development and delivery."

4) Built-in Integration with Web Services and Databases.

"Applications need to leverage existing software investments in databases, and internal or external third party web services, requiring that the platform offer a wide variety of connectivity options."

5) Support Collaboration

"A PaaS must support both formal and on-demand collaboration throughout the entire software lifecycle (development, testing, documentation and operations), while maintaining security of source code and associated intellectual property."

6) Deep Application Instrumentation

"With instrumentation, organizations can see exactly how users are using the application, the type of performance they are experiencing and any application crashes. This information can also be leveraged to create new business models where costs are tied to actual utilities, rather than flat-rate subscriptions or licenses."

Over the next couple of years we expect to be hearing a lot more about PaaS and how "Y announcement" by "X company" is now providing true a PaaS offering to businesses and developers.

But saying you are providing a Platform as a Service has to mean something, and we think the above definition sets a high but reasonable standard that must be met for any company to claim they are providing a "platform-as-a-service' and legitimately describe themselves as a PaaS player.

The amazing thing is, for me at least, is that Bungee Connect is delivering all of the above, today. From our point of view, delivering PaaS - the real deal - is not statement of Bungee's intent, it's a statement of fact. It's bold, but so is our vision. Yes, we've still a lot to do before we're commercially ready and we think that's coming soon, but so much is already there. Try it out.

8 Trends in Software as a Service Platforms

alexbarnett — Sun, 13 Jan 2008 17:40:00 GMT

To kick off the new year, I presented to around 40 or 50 members of Utah Technology Council (UTC) last week. The title of the topic they asked me to speak about was "Trends in Software as a Service Platforms". I searched around for some ideas and came across two recent posts predicting trends in SaaS for 2008, one by Phil Wainewright "Eight Reasons SaaS Will Surge in 2008" and Jeff Kaplan's post "Top Ten Reasons Why On-Demand Services in 2008". I decided to borrow liberally from these (thanks Phil and Jeff) and mash these two together (along with a couple of thoughts of my own) and present "8 Trends in Software as a Service Platforms" to an audience made up of CTOs and VPs of engineering and development for software companies in the Utah area.

In preparation for the presentation, my boss (Martin Plaehn) at Bungee Labs suggested I write up my presentation as notes blog them afterward, so here they are.

8 Trends in Software as a Service Platforms

SaaS is just part of the web mega-trend
Mainstream opinion says “Yes” to SaaS
Software vendors stampede into SaaS
All is being virtualized
Explosion of Web APIs
Economic factors favor SaaS
Enterprise and SMB IT embraces SaaS
SaaS platforms proliferate (PaaS)

1. SaaS is just part of the web mega-trend

Most of us have witnessed and many of us have been a part of the transformation in the way goods and services have been digitized, virtualized, delivered and consumed. Software, the data behind that software and the functionality that software provides is no different - software is subject to the very same transformational forces.

Just think about how even a class of product that is natively digital - such as software - has been transformed in the way it is delivered and consumed. For prosperity's sake, I've still got a few of those ZX81 software cassettes stashed away somewhere, gathering dust, looking ever more antiquated with each passing year. How will today's mode of software delivery and use look to us in a few years from now?

The web wants to connect things, and that's interesting. But connecting and interacting with "live" data, information and remote functionality make things more interesting.

At the fundamental level, the web connects things. It connects people to people, businesses to businesses, and people to businesses. Since the early 90's, the web has enabled the connection of so many things to so many other things at an ever accelerating rate, and yet we crave even more connectivity. But we increasingly also want the ability to interact with those things.

And it is the nature of these connected things that have changed since the early internet. The early web was good at connecting to static views of information and accessing limited and rigid functional services, very much a read-only mode. Then, as we learned a) the ability to read more dynamic-type information - at least regularly updated, and b) access richer remote functionality, we created whole new opportunities for ourselves. Next, we grew our ability read and write against dynamic, near real-time data and information and to program against remote functionality to create a new class of web applications leveraging those capabilities - and hence a new order of business and experiential opportunities have emerged. Some label this as "Web 2.0".

At its essence, it is the "liveness" of these real-time read-write data, information and functional sources available as "always on" services and the increasing ease to connect to, interact with - specifically change those resources available as live, programmable services that allows us to create new value out of those resources, opening up brand new market opportunities for businesses and the compelling, rich "live" end-user experiences of tomorrow.

2. Mainstream opinion says “Yes” to SaaS

Not surprisingly, Wall Street loves the the predictability of subscription services. It's good for cash flow, forecasting and business planning.

The venture firms also relish the opportunities that are opening up in a software as services-oriented economy. The ability to circumnavigate the incumbent software players with new disruptive technologies and propositions that are significantly easier to try and access for prospective customers compared to traditional software evaluation, along with usage and subscription-based business models verses the old licensing model makes investing in services-based software companies very compelling propositions from the venture firms' point of view. We should also see healthy M&A activity based on these similar opportunities in the coming year.

And then there's the trend for offshore / IT business process outsourcing. These providers will surely get in the game and make their plays through investments in and acquisitions of SaaS vendors that align well with their current core businesses.

Add to that the excitement we're reading about the SaaS space from the IT Analysts, journalists and bloggers, plus the new book by Nick Carr (author of “IT Doesn’t Matter”) - delivered by Amazon to me last week: “The Big Switch: Rewiring the World, from Edison to Google”. I think there's little doubt Carr's excellent analysis of the computing industry as an analogy to the electricity industry's shift to a utility model will be on business bestseller list for much of 2008. His messages resonates with corporate executives and end-users agree with him:

IT is a needless hassle,
it should be as easy as electricity and
be as reliable as a utility

3. Software vendors stampede into SaaS

The Big Software Players are following the early SaaS successes

CRM as a case in point. If you've been following the CRM software market, you'll know about the noises Oracle-Siebel, SAP and Microsoft started to make in the 2007 about what they are are lining up for the 2008 in terms of CRM as a service. Their efforts to emulate Salesforce.com's success delivering CRM as SaaS will be key strategic bets from the incumbents' point of view - and loud, price and functionally competitive propositions from the point of view of their existing and prospective customers.

CRM is just one of the multiple horizontal solution categories to transform from on-premise with traditional licensing model to a service-based delivery and subscription-based revenue model. ERP, supply chain, e-commerce, HR and many more...the horizontal solution list goes on. And then there are the vertical solution players...

Here's another data point to consider regarding the move by traditional software vendors to a SaaS model:

“15-20% of application ISVs have already either begun new skunk works initiatives or gained access to SaaS assets and development experience through M&A activity”
(Source: Key Trends in SaaS: 2008 and Beyond, Saugatuck Technology)

4. All is being virtualized

Virtualization is a technology trend.

Virtualization enables hardware as a service. The demand for virtual machines met by hypervisor software (VMWare, Xen, Hyper-V) and the success of Amazon's Elastic Compute Cloud (EC2) in the last couple of years point to a continuation of further virtualization of applications and hardware. Virtualization is accelerating the move from traditional on-premise software to services.

Virtualization is a business trend.

We continue to become a mobile workforce. The younger entrants into the workforce in service-oriented economies expect and want to be always connected. It's very hard work, if not impossible to get your traditional on-premise applications and centralized servers sitting behind a firewall to serve today's mobile workers. SaaS and managed services meet the needs square on.

5. The explosion of Web APIs is upon us

According to ProgrammableWeb.com, there are 559 commercial and public APIs available today, most of these are new and there are plenty more to come. How many will we see go live this year? And how many private web APIs are there and will be developed and consumed in the coming year?

Data from ProgrammableWeb.com

6. Economic factors favor SaaS

On-premise software requires upfront capital investments
To lower costs, many companies hold back on their capital investments to mitigate their risks, especially in recessions
Adopting on-demand services on a pay-as-you-go basis will be a perfect sourcing strategy for businesses seeking greater cost-controls and flexibility – the utility model

All well and good, but the real economic value of SaaS is that fact that it unleashes new value of previously isolated data silos and functionality.

7. Enterprise and SMB embraces SaaS

When it comes to IT, who doesn't like

Low-maintenance?
Low cost?
Low-resource profile?

IT and business folk like these things, and externally delivered SaaS applications deliver these benefits.

8. SaaS platforms proliferate (PaaS)

The more mainstream SaaS becomes the more the large vendors will be forced to offer effective platforms for ISVs, enterprises and SMBs.

If the move by the software vendors from traditional on-premise software to a services model is to be successful, they will need to provide programmable interfaces - not just end-user interfaces - to their services for their customers. Customers need and want the ability to access, intergrate and create new value out of live, programmable data, information and functionality living in the cloud. And in turn these same customers will want their custom-developed composite applications and integrated data available as programmable services - yet more APIs.

Customers want to unleash new value of previously isolated data silos and functionality through the development of their own applications programmed against those resources. And in turn these same customers will want their own custom-developed composite applications and newly integrated data available as end-user interfaces and as programmable services - yet more APIs. These customer needs will drive the software market to provide platforms to provide businesses and developers with with end-to-end:

programmable services and data integration
application development, testing and collaboration tools
deployment and scalable delivery

...all as a service with a utility model.

(hey...I needed to mentioned Bungee Connect just the once ;-).

2008 will mark a the proliferation of such offerings as "platforms as services" (or PaaS) through 2009, where then the consolidation will begin. Interesting SaaS and PaaS times ahead.

Update 2/20/2008: see "Time to Define "Platform as as Service" (PaaS)

The presentation seemed to go down pretty well and we had lots of interesting discussion throughtout. One of the topics we discussed was data security in a SaaS world. Don Kleinschnitz (VP, Development at Symantec) followed up with a mail linking to his blog covering Security 2.0 topics.

Again - thanks to Phil Wainewright and Jeff Kaplan for their post and to Martin for suggesting I blog this.