Can Joyent Create a PaaS Platform More Cost-effective than Amazon EC2?

At last week's Enterprise 2.0 Conference in Boston, I had the opportunity to talk to James Duncan, Director of Platform Strategies at Joyent. Joyent is a cloud computing vendor in the IaaS (infrastructure-as-a-service) market. They sell virtualized servers called Joyent Acclerators, which customers use for running Web applications.

The idea behind IaaS is that customers can grow or shrink their infrastructure dynamically. Expecting a surge of business, or need to crunch vast amounts of data for a rush project? No problem. IaaS lets you instantly access as many servers as you need: 50, 500, even 5000 or more. Project over? Again, no problem. Shut down the servers at once. You've managed to meet your computing needs without buying, configuring, and managing racks of servers yourself.

In January, Joyent bought Reasonably Smart, a cloud computing start-up that James and his business partner Bryan Bogensberger had founded the previous summer. First on their own and now at Joyent, James and Bryan have been developing the Smart Platform, a PaaS solution that lets programmers develop server applications in JavaScript and then deploy these applications to the cloud using Git, an open source version-control system. Now branded as a Joyent solution, the Smart Platform is ready to go into a private Beta test.

I asked James why he chose JavaScript as the programming language for the new platform. Why choose a language famous for improving interactivity in a browser, instead of a language more commonly associated with server-side applications, such as Python or Perl or Java?

He pointed out that JavaScript was designed from the outset to deliver a lot of functionality in a secure, self-contained environment (known in programming as a "sandbox"). Secure autonomy is a great feature for applications and services running in a shared environment such as a data center, where virtual servers from several customers might be running on the same machine. He also pointed out that more money is going into training and documentation work on JavaScript than into other popular languages, such as Ruby, PHP, and Python. And, of course, JavaScript is a proven language, already popular with lots of programmers, who have been using it for years to develop sophisticated client-side Web applications. Now these programmers can apply their knowledge of JavaScript to the development of fast, efficient server applications that run in the cloud.

The Smart Platform offers a couple of other advantages that cloud computing users may find attractive.

First, the platform is open source. Joyent hopes that customers will like Joyent's PaaS services so much, they'll subscribe to Joyent and pay them to run the platform. But customers and even non-customers are free to move the entire stack in-house. Running the platform in house could be helpful for developing prototypes, running private clouds (shared, elastic services run entirely behind the firewall), or combining private and public clouds, as needed. Also, by making the platform open source, Joyent expects to ease customers' concerns about being locked in to a particular PaaS vendor. (Vendor lock-in has been much discussed lately in cloud computing circles.)

Another important advantage of the Smart Platform is that it's truly priced like a utility: customers pay only when an application is servicing requests. They pay nothing when is present on a virtual server, but inactive (that is, not processing data or interacting with other applications). This utility pricing model has the potential to make Joyent far more cost-effective than even Amazon's highly affordable Elastic Cloud Compute (EC2) service.

James offered this example. Let's say you're hosting your Web site in the cloud for a month. On Amazon, you would pay for the Amazon Machine Image (the Amazon virtual server) for the entire month, even if the Web site gets only a few hits and actually active for a few seconds or minutes.

On the Joyent Smart Platform, in contrast, you would pay only for those few seconds or minutes of activity; you would pay nothing for all the time the server is hosted on the Smart Platform inactive. As a result, your bill should be a small fraction of what it would be from Amazon.

Pricing based on activity rather than deployment could make a big difference for customers whose hosted applications are active only sporadically. It could also be attractive for customers who have forgotten to decommission servers once a project is over and have then been surprised to receive a big bill from a cloud service provider.

If you're interested in participating Beta of the Joyent Smart Platform, visit www.reasonablysmart.com.

How Good Is The Interface to Your Web Site or SaaS Application? Find Out Now

I've written before about the importance of customer experiences. When I look back on the companies I've worked with over a couple of decades in high tech, the quality of experience a product provided largely determined the success or failure of the product overall.

In the software market, user interfaces obviously play a major role in a customer's experience and, hence, in a product's success or failure. Complex, cumbersome interfaces confused users, prolonged sales cycles, and often limited buyers to sophisticated engineers working in solitary roles. Streamlined, easy-to-use interfaces, often developed in concert with customers, led to faster sales, more efficient operations, and greater business success overall.

There's a wealth of material available on the importance of user interfaces and customer-driven design. Jakob Nielsen, Jared Spool, and Steve Blank all have important things to say on these topics.

Of course, if you're in the middle of a product launch, or you're a product manager working to an aggressive schedule, or you're a programmer sweating to make your deadlines, you probably don't have a lot of time on your hands to do in-depth reading and learn a new set of best practices.

Nonetheless, you'd probably like to make sure the Web site or software you're building meets the needs of your customers to the greatest degree possible, since this is the most direct and cost-effective way to:

• ensure rapid adoption by customers
  
• shorten sales cycles
  
• reduce customer support calls
  

So, busy with development and probably short on time and money, how can you ensure you've got the best user interface possible? How can you nip a problem in the bud now, before it turns into a full-fledged customer support nightmare later?

Introducing Mad*Pow

You can take advantage of a new service I'm now offering with my new business partner, Mad*Pow.

I didn't want to merely write about improving customer experiences, wringing my hands without offering a concrete solution. Instead, I wanted to be able to help any interested companies address this problem quickly and directly. So I met with usability designers, explained what I was looking for, and ended up forming a partnership with Mad*Pow, a leading experience design agency.

They're great folks. They're based in Portsmouth, NH, and their clients include:

• Aetna
  
• American Express
  
• ESPN
  
• Google
  
• Intuit
  
• Journal of New England Medicine
  
• NBC Universal
  
• Starwood
  

You can check out some of their projects here.

Incidentally, Mass High Tech just named Amy Cueva, Mad*Pow's Founder and Chief Experience Officer one of its "Women to Watch" award winners.

The company was founded in 2000, and, as you can see from the client list above and from their Web site, they've clearly grown into a dynamic, successful company. I'm thrilled to be working with them to address the user experience problem.

Rapid Strategic Analysis of User Experiences

So what does our new service entail?

First, we’ll interview you to understand your business objectives and design goals. Then we’ll conduct a thorough assessment of your Web site or Web application. Our analysis will include:

• Heuristic analysis by multiple testers and designers to uncover usability weaknesses.
  
• Competitive analysis (2-3 competitive or comparative sites)
  
• Marketing and messaging analysis
  

We will then deliver an annotated slide deck and a written report detailing our findings, including:

• Analysis of content, navigation, design, brand, and messaging
  
• Vision for the future—suggestions for improvements in all areas analyzed
  
• SWOT (strengths, weaknesses, opportunities, and threats) analysis summary of the user experience overall
  

Our report will answer these vital questions:

• Is your Web site or Web application’s user interface as easy to use as possible? Are any design elements confusing or cumbersome? How well does the interface serve its users?
  
• How well does the interface serve your various audiences?
  
• Does the interface support customers making full use of your offerings? Does it facilitate upgrades and cross-selling?
  
• How does the interface compare to those of your competitors?
  

In addition, you can ask Mad*Pow's design team to offer various design alternatives for your site or application.

We've designed this process to achieve quick turnaround: we expect to work from kick-off interview to delivered report in 5 business days. And we've priced it in fitting with the budgets that are prevailing in today's, um, wonderful economy.

You can read more about the service here. (PDF)

If you're interested in having our team kick the tires on what you're building, please contact us at experience @ bennettstrategy dot com.


Lightbulb photo by rpongsaj, some rights reserved.

Who's Using PaaS? The Answer May Surprise You

When Platform-as-a-Service (PaaS) vendor Coghead hit the skids recently, other PaaS vendors such as Intuit QuickBase and TeamDesk were quick to offer Coghead customers free conversion tools and migration solutions, so they could keep their applications running smoothly in the cloud.

I was curious about the customers making this move. Had they been they using Coghead for in-house skunk works, pilot projects, or operational applications? In large companies, did management concerns about data security, SOX compliance, and other regulations relegate PaaS to department-level projects that were, shall we say, off the management radar screen? Is PaaS (when not an extension of a proven SaaS solution such as Salesforce.com or Sugar On Demand) simply a way of getting code up and running without having to requisition a server from a bureaucratic IT department? Or is PaaS something more?

To find out, I emailed TeamDesk, asking about their customers and the types of applications they were running. I promptly received a phone call from Val Karmazin, co-founder of ForeSoft Corporation, the company that offers TeamDesk. Promptness, it turns out, is a habit at ForeSoft. The company prides itself on prompt, reliable customer service.

ForeSoft offers four cloud computing solutions:

• BUGtracka project management and issue-tracking application
  
• dbFLEX, a platform for building business Web applications
  
• CRMdesk, a help desk application
  
• TeamDesk, a platform for building and easily configuring database applications, primarily for the back office
  

The company was founded in 2001. It's been profitable from the start and hasn't taken any outside investment. TeamDesk, launched three years ago, is now the fastest growing part of the business.

In our conversation, Val made a number of interesting points:

• The TeamDesk user community is a mix of people, and most of them aren't developers. What? I thought PaaS would appeal mostly to developers who were comfortable with Ruby, Python, Java, virtualization, and so on. That's not so in the case of TeamDesk. The platform is so easy to use and so easy to configure, thanks to a configuration dashboard, that many of the users are small business owners or IT engineers who know enough about databases to establish a relationship between two tables, but not much technical knowledge beyond that.
  
• Companies of all sizes are using the service. In addition to small businesses, Val named a major telecommunications carrier, a major shipping company, and a European office of a major Silicon Valley technology provider. Company size doesn't predict whether or not a customer will use PaaS; instead, the determining factor is company culture—how willing is the company to trust a PaaS vendor to do things right. Val says customers take advantage of TeamDesk's free trial, discover how easy to use and reliable the service is, and stick with it.
  
• Val points out the distinction between PaaS and SaaS is often illusory. When his customers use his platform to develop business applications that they then rely on day after day, haven't they created in effect a SaaS solution? Read TeamDesk's case study about the sales management application that Hochkoeppler Initiatives created for a customer, and you'll see what he means.
  
• TeamDesk users sometimes require a little more hands-on assistance from ForeSoft when they're getting started with the service, but so far the support workload remains manageable.
  

A couple of conclusions:

• PaaS is ready for prime time: customers are using it for business applications, not just for development projects.
  
• With the proper tools and application user interface, a PaaS vendor can reach beyond the development community to less technical users, broadening the pool of potential customers.
  
• Small, focused cloud computing vendors who execute well can run a profitable business.
  

Coghead's Demise is a Reminder to Sober Up about Cloud Computing's Promise

I was sorry to learn that Coghead, a Web 2.0 Platform-as-a-Service (PaaS) provider, announced it was shutting down.

PaaS is an interesting model: a company offers a hosted service for developing, testing, running, and monitoring new applications. Customers can use the PaaS platform to launch new applications—or scale up existing ones—without deploying any local hardware or software at all. Sounds intriguing. Possibly very convenient. Possibly cost-effective.

It's proving to be too futuristic a vision. First, Bungee Labs, another PaaS provider, ran into trouble in 2008. Now Coghead is shutting its doors.

You wouldn't think this would be possible to read all the hyperventilating blog posts about cloud computing: how the time is right for cloud computing, everything will run in the cloud, what is the cloud—do we include PaaS? SaaS? If we don't define cloud computing properly, the cloud will perish! Many industry insiders are in a lather about these issues. It's as though the lottery has come up with the winning letters, which spell cloud computing, and now we have to scratch off the bonus letters just right to multiply our winnings or be sent home with a PC, jr. Nerves are a-jangle. Fingers are flying.

But if cloud computing is such an obvious remedy to the IT woes of business, why are these vendors in trouble?

Let's come down to earth for a moment. Let's consider this situation from the business customer's point of view. The business manager asks, Do I need PaaS? Is this really the most convenient, least risky way of building and deploying new applications?

I don't want to sound pedantic, but I think many cloud computing services will have trouble overcoming the obstacle that's snared many other technically impressive solutions in other IT markets of yore: just because you can build it and it's cool, doesn't mean that business customers will be comfortable with it. How many business customers in this increasingly risk-adverse environment are really going to adopt new programming paradigms and hosted services from small, evidently risky providers? Not many, I'm afraid.

Businesses want low cost and convenience. They want reliability and low risk, just as much. And (almost) nobody in business likes to learn anything new unless they have to.

I do think there are attractive opportunities for cloud computing in 2009. I'll details those in my new post. (Don't worry. It will be short.)

Note: Thanks to @chris_marino for tipping me off to the sad news about Coghead.