According to Gartner, Software as a Service (SaaS) is software that is owned, delivered and managed remotely by one or more providers. This means that the application users are not licensed and charged for software availability in extended periods of time, but only billed for the amount they actually use. In most scenarios, the software is either available in the form of web applications or terminal services. In the first case, the entire application is hosted on the provider’s hardware and no client software except for a web browser is needed. In the latter case, the only difference is a requirement for the customers to download a client application, but the core of the system is also hosted by the provider. Benefits from SaaS These facts combined mean vast savings for the consumers. The lack of an initial license fee and hardware requirements can reduce the CAPEX significantly. It is also easier to plan the spending and adapt over time. The actual cost of ownership (TCO) depends on how much the applications are used at a particular time and not on future capacity. This flexibility and affordability of the model are especially vital for businesses in today’s economy. As for the software vendors, the SaaS model offers equally valuable benefits. Initially lower, but recurring revenue streams are much more predictable and provide the ability to plan the budgets more effectively and precisely. Due to the centralized hosting, the software is also much easier to maintain and support. All upgrades are limited to one environment and have instant effect for all users. In addition, direct access to the application logs facilitates bug fixes. Finally, SaaS can help overcome sales difficulties in a period when businesses reorganize and freeze their IT budgets, so they cannot afford the lack of flexibility and expenses of software based on EULA licensing models. These advantages are clearly confirmed by good results of the market leaders and optimistic projections of its researchers. Contrary to mostly negative growth forecasts coming from all over the economy, the global SaaS market is expected to grow in 2009 by as much as 30% (Gartner) to 40% (IDC). Billing and other challenges The positive aspects of SaaS for software vendors are unquestionable. However, a number of topics need to be addressed before an application can be offered in this model. The inevitable challenge faced by all Software as a Service providers is setting up the billing process. Whereas traditional IPR or EULA-based sales required simple license invoicing and handling of usually long-term maintenance contracts, the “pay-per-use” model and proper management of frequently recurring transactions impose a requirement for a rating and billing engine, as well as a set of procedures. This means additional analyses and investments need to be made in order to kick off the provision of SaaS. The necessary infrastructure is offered by many vendors (e.g. Verax Systems with its OSS/BSS Billing). In order to achieve good results, software businesses are required to develop a profitable and competitive usage billing model. One of the first steps is defining the main billing units and UDRs (Usage Data Records) related with them or software license key limitations. The most commonly used aspects are: Number of users and sessions per user Number of concurrent sessions Number of enabled modules / functionalities Number of business artifacts generated by the application (e.g. reports, invoices etc.) Number of objects created or stored in the application (e.g. articles, contacts etc.) Number of emails sent Obviously, the rating and billing must cater for the business value of the applications, service maintenance costs (like customer support and SLAs), as well as the hardware required to host it (e.g. CPU and storage capacity). The diversity of the parameters may be a difficulty alone. However, this is where another critical challenge occurs. It is the scalability required to handle a varying number of customers and users. Obviously, a well-established business can make long-term customer base growth plans and set sales targets in order to adapt the infrastructure on time. However, the recent economic reality has made it increasingly difficult for companies to reach those targets. In addition, some of the services offered to customers have a very seasonal nature (e.g. consumer e-commerce usually booms in the Christmas season). This means businesses need to make upfront spending on hardware capacity which is likely to be redundant for extended periods of time. A related challenge is also the provisioning of the services, which also requires appropriate infrastructural solutions to be in place. A conclusion from the above is that it is not easy for a specialized application provider to offer their software in the SaaS model on their own. Fortunately, the market is rich in solutions similar in the idea, but oriented on hardware infrastructure. It is usually referred to as Infrastructure (or Platform) as a Service, and a combination of the services is commonly named Cloud Computing. “Hardware as a Service” The Infrastructure as a Service providers reduce most of the CAPEX required from software vendors in order to start offering SaaS. Their huge data centers cater for the flexibility allowing for instant multiplication of the hardware resources as the needs grow. The dynamic scalability and provisioning is achieved with the latest platform virtualization monitoring infrastructure (hypervisors), out of which the most commonly used are Cytrix Xen and VMWare (Information Week Analytics, Sept. 4, 2009). Costs are kept down to the minimum due to built-in load balancing mechanisms. The dynamic growth of interest in SaaS had turned providing scalability, redundancy and provisioning for its purposes into a core business of many companies. Even though, as the concept is relatively new, the implementations and market offerings differ quite considerably. The most commonly listed three services – Amazon’s EC2, Google’s App Engine and Microsoft’s Azure represent different philosophies, with hardly any platform restrictions and added services in the first case, very restrictive policies for a low price in the second, and single platform with value added services in the last example. With specializations ranging from virtualized and scalable web hosting and disaster recovery through provision of SaaS and test environments for software vendors to leasing high-performance computing resources for research and industrial simulations, the leaders in the most common appliances include Amazon (EC2), Rackspace and GoGrid. A majority of the providers impose a minimum service duration, although in most cases it is as low as monthly. The services are usually billed according to utility-based or availability models. The charges are commonly applied for the following parameters: Hours of virtual machine availability (e.g. Amazon) CPU cycles (e.g. Rackspace, Google) RAM-hours (e.g. GoGrid) Data transfer Storage Additional services, such as monitoring, load balancing, software license fees etc. are also offered and billed for as part of bundled plans or separately. Some providers offer pre-paid plans and monthly or annual subscriptions, although their practical aspect is a price discount or a fee for the “reservation” of a machine (either virtual or physical) with additional per-use pricing on top of it. One of the most frequently raised disadvantages of entrusting the hosting of applications to 3rd party companies is the aspect of data security and uptimes. This is addressed by most providers who offer suitable service level agreements (SLA) with uptime levels exceeding 99%. However, it is the small-print that matters. For example, Amazon’s SLA guarantee of 99.95% is calculated on an annual basis, which means a critical system may be down for a few hours within a week with no obligation from the provider. As another one, GoGrid’s 100% SLA level refers to availability as indicated by the operator’s proprietary monitoring tools. Service separation model In this model, the software vendor hosts its applications in a selected Data Center providing platform or infrastructure services. The software is offered and sold to end users directly by the application provider and the data center is not part of the process. The application provider is billed for the infrastructure usage. The application sends usage reports to the provider’s billing engine. The entire billing and invoicing process is also handled by the application provider. The main advantage of this model for the application provider is that the scalability and provisioning is entirely taken care of by the data center. This means a significant cost reduction, as no hardware needs to be purchased and set up in order to provide the service. The sales is directly between the software vendor and the customers. Both the data center and the application provider offer their core business services only. Despite offering undoubted advantages, this model is not without flaws from the software vendor’s perspective. The requirement of running dedicated sales & marketing departments has been enough of a struggle for many software engineering businesses. The billing and invoicing on top of that may be too much for some executives to handle in a short timeframe. Revenue sharing model This is why an alternative and less conservative model is proposed by Verax Systems. It is based on the assumption that it is easier for a [...]