Microsoft is pulling out as many stops as it can to position itself as a leading contender in the burgeoning Software as a Service (SaaS) marketplace, not least with events such as the latest Architects Forum held this week in its Reading fastness.

The Small to Medium-sized Business (SMB) sector is very much in the company's sights, and its partner channel is the weapon it is planning to use to grab a share of the business.

"The trend now is that SMBs don't want their own IT, they just want access to services," Gurprit Singh, director of emerging technologies developer and platform group Microsoft UK, told the assembled architects.

"But there is still a need for flexibility, with services hosted onsite or remotely and perhaps running a mix of the two."

He cited Exchange as an example of this, in that it can be hosted onsite or in the cloud, and can attach externally hosted services as required.

According to Matt Deacon, chief architectural advisor for the developer and platform group, this issue can be seen as part of the general problem of managing innovation for ISVs, particularly where SaaS as a service delivery vehicle starts to make a greater impact as a disruptive technology.

Office as a services composition vehicle

Deacon says: "Office has, of course, been a mechanism for bringing together applications functionality since the days of COM. We are just extending that idea to include online external services, such as SharePoint."

Said like that, he seems to suggest it is a purely Microsoft-centric capability, but he denied this, saying it could work with any web service, with Office being used to "customise the user experience".

Deacon highlighted some of the advantages that SaaS brings to the SMB market that the partners could exploit, such as Pay as you Grow, a fast start, little or no user training required, and vastly reduced IT capital expenditure.

Benefits for the ISVs include the fact that it does give them better application lifecycle control, as well as important business advantages - such as far easier access to the "long tail" marketplace of the small and niche market customers where the cost of sale means they are generally not worth pursuing.

A very practical overview of why ISVs should consider operating in the SaaS/on-demand marketplace was given by Microsoft Architect Eric Nelson, and the list of reasons is quite long.

For example, there is the chance to gain predictable revenue, which is now often referred to as the annuity business model where revenue is a constant drip-feed that can be planned for far better than the traditional feast/famine model. It also gives greater agility to meet new requirements when entering new markets and access to new revenue sources through the ability to sell to the "long tail" of small customers.

The ability to plan revenue better can lead to more, or at least better, investments in R&D, while the reduced requirement for customisation can dramatically cut sales cycle times and sales costs. "Once you say, 'we can customise this for you', customers will ask for 4,000 changes and you are straight into protracted negotiations, complex testing, and real problems with future upgrades that are not compatible with past customisations."

The alternative is to make applications easily configurable while leaving the core code alone.

Maintenance and support costs are reduced because fixing a bug for one customer automatically fixes it for all, even if they haven't come across the problem yet. If a real need for some beneficial customisation is found it can be put into the core code for the benefit of all customers.

Lastly, SaaS gives a good opportunity to build greater customer loyalty through the ability to respond faster to customer needs - and sometimes have the response incorporated before they enquire about it. There is also the opportunity to respond faster to competitive offerings. One of the best levers on customer loyalty, he suggested, was the ability SaaS gave for the try-before-you-buy business model.

The architect's view

From the architect's perspective, there are a number of issues to overcome in building a workable SaaS environment, of course. One of the key problems is the way databases are organised within the system.

"Nirvana would be running a single database with lots of tenants and all their data in the same tables," he said. "But a multi-tenant database is hard to do."

There are, for example, many management issues to consider, such as Service Level Agreements. "What if one tenant manages to hog the majority of the resources available on the single server?" Nelson asked. "There are also regulatory and governance issues where it might occur that data is shared between different tenants." This can, of course, be approved by the tenants, but without sufficient management resources and expertise, it could just easily be, at best, "inadvertent".

The alternative is to have either a single node and database for each tenant, or a single node running a multi-tenanted environment where each tenant had a separate database. According to Nelson, the practical limit here is around 200 databases per node and, in both of these alternatives, it does mean upgrades need to be made to each database individually.

Configuration issues can normally be restricted to the presentation of the customer's UI and associated branding, setting up the required workflow and rules, and the creation of data model extensions where necessary.

To help architects and developers get an idea of how to manage such tasks, Microsoft has introduced a dummy human resources "service" called LitwareHR. Here they can not only play at configuring the service to fit their requirements, but also examine the processes behind the changes in some detail in order to learn how to build applications that can be readily configured.

One way of getting a feel for providing online and on-demand services is now being offered by Microsoft's chum, BT. According to Joe Black, BT's director of emerging business and technology, the company has already invested around £8m in building the Connected Services Framework in collaboration with Microsoft, and one of the main reasons is that BT's own research estimates that the UK SaaS marketplace will reach £144m during 2009. Black suggested, however, that BT faces some practical problems in dealing with ISVs looking to get a share of that business.

"One of the main problems is identifying real demand when ISVs come to call," he said. "BT needs some form of market reference from ISVs to show that there is a demand for what they are offering, so there is an issue of how they engage with BT."

One solution he proposed is that ISVs use the Application Marketplace that has been established as a community in BT's Tradespace service specifically as a place for ISVs to get a foothold in the online world.

Black also pointed architects at BT's Web21C SDK development toolset. This provides a set of libraries designed to help developers consume web services that are exposed by BT. The idea is to give developers a simple line of code to access each one of the many communications-related services BT now has available.

"If you think an application can exploit a phone or communications service one line of code is all that is needed to integrate it," he said.

BT has also got involved with Microsoft in the development of the latter's Connected Services Sandbox where ISVs and other collaborators can build and try out new developments in a safe environment.

For those ISVs that feel they have a product or service worth exploiting, but don't want to take the step into the SaaS marketplace, BT is also in the market to provide that service.

Black said the company wants to co-market such products and services through a new ISV partner programme. This could lead to BT-branded products appearing. "This would be a revenue share model with the partners, with the share probably being 60:40 in BT's favour," Black said with a smile. ®