You don’t have to be part of the clouderati to know that SaaS (Software as a Service) does not necessarily mean cloud computing. And yet, for many, especially in the business world, SaaS is part of cloud. Though the specialists amongst us know this to be inaccurate, the question is “should we care?”. It depends on who’s answering the question.Any definition of cloud computing (eg here’s one from the NIST) will usually include the following essential characteristics:
- A rental model based on actual usage
- Standardized computing services
- On-demand self-service, with the ability to scale usage up or down
- Multi-tenancy, ie pooling computer resources to serve multiple customers
- Automatic service provisioning, which results in an elasticity which creates the illusion of infinite capacity
Such a model naturally favours volume and scale, resulting in higher resource productivity and a more efficient use of capital, and hence lower unit costs. Typical examples of cloud SaaS that adhere to all of the above criteria include Google Apps, Amazon Web Services and Facebook, which can sign up new customers automatically without any human interaction to process requests, set up accounts, provision servers and manage spikes in demand which require additional capacity.
Then there are other SaaS applications that only adhere to the first criterion, and only sometimes to the second and third. While some of them might also lay claim to multi-tenancy, few or none do automatic service provisioning. At the one end of the scale, such SaaS applications are nothing more than “off-premises” or hosted versions of on-premises software, and at the other end we can have advanced multi-tenant features. Examples are B2B vendors in the field of CRM, ERP or PPM, some of whom offer the same product in both the traditional on-premises model and in SaaS.
The main reasons for the absence of multi-tenancy and automated provisioning are twofold. Firstly, the volume and predictability of usage would not be able to cost-justify either. For example, a CRM customer is not going to increase his sales force by 25% overnight. And PPM users in an IT department are not suddenly going to unpredictably vary the way they use the application and generate a 50% increase in capacity within the space of a week. Secondly, the level of complexity in business applications are such that they need to go through an intermediate project phase which covers mapping the customer’s business requirements to the tool’s configuration options, a process that can take anywhere from a few weeks to a few months.
Clearly, in many if not most of such cases, multi-tenancy and automated provisioning cannot be cost-justified. Both are complex and expensive undertakings, whose payback can only be achieved through volume and scale – hardly the criteria that characterizes your typical B2B environment. (For an idea of the complexity in building multi-tenancy, check out the following article, Why Multi-Tenancy is Key to Successful SaaS).
Should this be any reason not to call this type of SaaS cloud computing? From a customer perspective, I’d say no. Your average SaaS business customer doesn’t know the definition of “true blue” cloud computing anyway, and probably couldn’t care less. From their perspective, cloud and SaaS are one and the same thing, which they would probably define as “a rental-based model that provides on-demand internet access to applications or services owned and managed by someone else” (read not by my IT department). The level of adherence to true blue cloud criteria will make itself felt in the form of practical things like costs, lead times or service levels. (Some would also add to that list vendor lock-in, but the jury’s still out on that one).
If the business world sees cloud and SaaS as one and the same thing, then many such SaaS vendors will understandably “cloud-wash” their offerings (ie latch on to the cloud label). Doing so makes business sense and allows the market to better identify SaaS vendors. When you come to think of it, there is nothing inherently deceptive about cloud-washing (at least when it comes to SaaS) – even I use the term “SaaS cloud” on my website to reflect this market reality, and I don’t get hung up about it.
To take a simple analogy, if I call up a taxi company to pick me up at 6am tomorrow morning to go to the airport, I might end up with a private car or limo rental which charges a flat rate, or a taxi with metered charges. And if I call another taxi company during the day to come and pick me up in 30 min, I’ll probably end up with a metered taxi. But to me they’re all taxis, whether charging is metered or flat-rate, and regardless of whether they’re “provisioned” from a huge fleet of cars which gives me the illusion of infinite capacity, or whether it’s one guy’s private car. For me as a customer, I only care about costs and quality of service, and I make my choice accordingly. And, guess what – the private car or limo company is probably also listed under taxis in the yellow pages.
For a computing analogy, when Windows 2 came out in the late 80s, Apple afficionados – and indeed some people at Apple itself (I used to work there) – started saying that said it would never take off because it wasn’t a “real GUI (Graphical User Interface), built from the ground up”, but simply a “paint-job sitting atop MS-DOS”. We all know where that argument went. Customers made their decisions based mainly on costs (cheaper than Mac) and quality (clearly inferior, but nonetheless acceptable).
In conclusion, business leaders with business problems to solve will turn to “SaaS” or “cloud” or “SaaS cloud”, and we should not let the distinctions get in the way, because they will base their decisions mainly on costs, lead-times and quality of service.