Killer Applications in the Cloud! That was the theme of a pretty exciting panel
discussion at IEEE Cloud 2010, organized in Miami, FL during the week of July 5th. What is a killer app? In the extreme, these are the sort of apps which so enthrall millions of users that they would willingly spend their time, attention and money for a chance to use them. The Web browser, games, multimedia, productivity applications, popular desktop and mobile operating systems are prominent examples of killer apps. They all have a few things in common: they enable use cases – key functionality – users need and want, they are accessible, are easy, even delightful to use and they add economic value. In addition, a true killer app also enables something fundamentally new, something unknown or difficult to do prior to its invention. Do we already have such killer apps on the Cloud? If not, why not, and when will they arrive? What are suitable apps for the Cloud, let alone killer apps? These are some of the questions in front of the panel – not very easy to answer. At the outset, we must also clarify what the Cloud itself is. Following the IEEE theme, we will take the Cloud as scalable services consumption and delivery platform enabled by a distributed, virtualized IT infrastructure, run by an autonomic management framework. To address this question, we need to divide the time and space in which Cloud happens into at least a few bins: short-term and long-term, personal apps and business apps, installed apps and web apps, for example.
In case of personal apps, the answers are clearer – isn’t the world already using social media powered by the Cloud? We already have more than 60 millions of blogs, Wikis, Facebook, Twitter, Myspace, Cyworld, Gmail, and the list is rapidly growing. Enterprise cloud apps are expected to follow suit. CRM apps, popularized by Salesforce.com, are an example of a compelling app provisioning model via SaaS: mobile sales force accessing, updating and managing their customer and sales resource information on the fly is a new, valuable use case. When it comes to the Enterprise in general, there are a wide variety of applications and user experiences, with significant existing investment, resources, knowledge and skills. For example, there are more than a million software applications built for Microsoft Windows, the primary OS used by the Enterprise and home users alike. There are a large number of applications running on Linux and Solaris as well, especially from the research and academic communities, which are of vital importance to the economy. Applications on the Macintosh OS have gained popularity due the great UX they offer. In all these cases, the apps must be installed on the OS, integrating their components such as the linked libraries and drivers through both the user and kernel modes of the OS and its registries. It is estimated that the apps following this traditional software installation model, known as the Locally Installable Apps (LIA), require up to 5 times their initial cost for their maintenance. The SaaS model offers to cut down these costs significantly. LIA have one unique usability and performance advantage: they have quick and easy access to the key computing resources such as memory, CPU and storage. They do have their own share of pain as well, such as the familiar app crashes and the blue screens of death. In contrast, applications which are installed on remote servers, such as the Rich Internet Applications (RIA), depend on the remote machine resources for the most part. There are an estimated 1 trillion unique URLs on the Web, which implies that there are at least several tens of millions of unique web apps. As we can recall from experience using the Web, performance of such apps could sometimes be degraded, due to limitations in resources such as network bandwidth and speed as well as remote server resources. ‘Server is currently down!’ is an error familiar to all net surfers. What SaaS promises to deliver is the richness of functionality and features traditionally offered only by the LIA, and the convenience, scalability and cost advantages of the RIA, packaged into Software as a Service.
Transformation from the existing IT to the Cloud is expected to be certain, but gradual. Enterprises cannot afford to have any disruption to the use of their existing IT systems and Line of Business (LoB) applications, and also cannot spend much time training their busy staff with the new Cloud platforms. In the short-term, then, the Enterprise will look for Cloud apps and technologies which will help make this transition smooth for their existing use cases, while welcoming novel Cloud apps which enable new use cases. Apps which facilitate the transition will be valuable in the short-term. For example, if an organization uses legacy web apps and locally installed apps together for organizational work flows such as HR, payroll processing, travel accounts, procurement etc., and cannot afford a complete revamp with a software replacement, the Cloud ‘app’ or platform which addresses this need in a seamless way would be a valuable app. When it comes to the Enterprise, there is seldom a single ‘killer’ app. Instead, there would be hundreds of flagship apps by each industry vertical, addressing specific needs. In the long-term - that is, when the key concerns such as security are addressed and the business world completes its transition into the Clouds and is ready to embrace the innovation – the true killer apps would begin to emerge.
Recently at Runaware, we packaged an oil reservoir engineering software on our cloud platform as SaaS. A reservoir engineer evaluating the heavy weight simulation software was pleased to see it presented as a web app, and commented that such tools are valuable especially in remote, offshore locations where an engineer can access high end simulation results in real time from their laptop and make decisions about exploration, without having to worry about hardware, resources or expertise to run such jobs. It then occurred to us that this same SaaS platform is also an inexpensive test bed for ISVs to launch pre-release and pre-production Beta testing versions for their software. It is well known that Beta testing helps ISVs significantly improve the quality of the shipped products by identifying and fixing critical bugs well ahead. However, majority of ISV software ship without wide Beta testing. Cloud platforms could address this need in a quick and easy way. These are just two new use cases which the Cloud model uniquely enables, which the traditional app platforms simply cannot. Layered applications, known as ‘mash ups’ are another case in point. Imaging GIS based apps serving different verticals, from different vendors: software for flood routing, for forecasting earthquakes, for transportation planning, emergency management and for disease control. In an emergency such as the hurricane Katrina, many such software, powered by the Cloud and delivered on demand over the Web, can be overlaid on a base regional map, to make well informed, real-time decisions to manage the emergency well, minimizing the losses to property and lives. Similarly, SaaS software by different vendors could be assembled together in real time, as components in a complex workflow to enable different Enterprise scenarios, used only as and when needed on demand, and dismantled later to make room for the next Cloud app! As the tight grip of the hardware and the OS on the apps loosens and as the mobile devices become popular computers, new generations of cloud apps will enable scenarios we are yet to imagine. The Lego Land of killer Cloud apps is so fun to imagine and wonder!
