Two and a half years ago at the Web 2.0 conference, I and a few market observers were asked what cloud computing was all about. Back then, the concept of cloud computing was just beginning to gain mindshare with valley-based startups but many weren’t sure what it really was and what it meant to consumers and businesses. Responses from industry veterans ranged from greatly simplified (elastic infrastructure) to utterly bizarre (liquid paper!). A lot has happened since, and the term ‘cloud’ has become one of the most misused pieces of jargon in the technology world today.
As it becomes increasingly evident that cloud computing is here to stay, PI takes a look at how the phenomenon has shaped up over the last two years, and more importantly, what it really is and what it isn’t. As most pundits seem to agree that SaaS, IaaS and PaaS are the three mutually exclusive and cumulatively exhaustive manifestations of cloud computing, we’ll explore each of these models in turn.
Software as a Service (SaaS)
The earliest avatar of cloud computing, SaaS gathered a lot of traction over the last two years, with Salesforce.com crossing $15bn in market value, and smaller success stories NetSuite and RightNow valued at over $1bn each. There were notable failures too – LucidEra, which had raised over $15mn in venture funding to deliver SaaS Business Intelligence (BI) solutions shut shop last year, as did Blink Logic, another SaaS BI provider. The failure to achieve mainstream success outside a few niche domains has failed to take the sheen off the SaaS model though, with almost every new business software venture these days invariably being a SaaS offering.
Having evolved from being a mere extension of the Application Service Provider (ASP) model to being a true multi-tenant architecture model, SaaS gained many advantages relative to traditional on-premise and packaged software, fuelling a spurt in market adoption. With Gartner pegging market growth at a whopping 16%, 2010 has definitely been a landmark year for SaaS. Even as the future looks rosy with annual growth rate expected to reach 20% by 2012, all isn’t well in the world of SaaS. Many overzealous marketing folks hoping to cash in on the popularity of SaaS are choosing to airbrush some very crucial differences between SaaS, hosted (ASP) and on-demand models, misdirecting their customers in the process and in some ways making SaaS a victim of its own success. Here’s the PI take on what SaaS really means.
What is and what isn’t SaaS?
Many advantages attributed to SaaS, such as affordability, scalability and flexibility, are direct consequences of true multi-tenancy, which essentially allows many different client organizations to access a single instance of an application running on a physical or a virtual server. Put simply, there is one and only one version of code across the entire customer base, removing the expense and effort involved in maintaining multiple versions or customer-specific code and managing different production environments.
In the hosted or ASP model, the service provider hosts different versions of an application for different clients in a centralized location, often hosting each version of code on a separate physical server. This
model does not guarantee full multi-tenancy and hence does not fully offer the advantages that SaaS does. However, this model is very suitable for applications that require a high degree of client-specific code customization.
On-demand software is a much broader term that encompasses both SaaS and ASP, and is only reflective of how the users are charged for access. Many large and well-known vendors often try to slip in traditional non-SaaS applications as SaaS to unsuspecting customers by tweaking their pricing model from per-server or per-site licenses to usage based metering (by number of users, transactions or storage capacity). While this pay-as-you-go model will technically make the application ‘on-demand’, it’s definitely not SaaS and will not be able to offer all the advantages that SaaS deployments bring to the table.
What are the implications?
If it’s SaaS, customers don’t own the application. Instead, they merely pay a service fees based on consumption to rent the application. This gives them the ability to test the waters without making any heavy upfront investments in the technology.
Customers pay on either a per-user or a per-transaction basis, and can expect near-instant and near-unlimited scalability. This gives them the freedom to start small and scale up without hassle to meet their business needs.
While customers normally can’t choose where their data is hosted (data isolation is only offered at the application level), they don’t need to worry about how and when to upgrade. All upgrades are carried out by the vendor and are typically non-disruptive, removing the need for tedious and expensive data migrations.
If it’s ASP, customers own the application after paying upfront for a license (this cost may sometimes be waived off, but is eventually funneled back to the customers in the form of maintenance fees). They can
expect to pay recurring annual fees for hosting services and general upkeep of the application. The fees maybe based on a per-user or a per-server basis. Customers normally can choose where their data is
hosted (physical isolation is possible instead of virtual isolation) and will likely be charged additional fees to be able to upgrade to the latest version.
A word of caution
If you are a startup or a business planning to buy a SaaS solution, make sure you quiz the vendor on the nature and maturity of multi-tenancy offered in the product. This is a good indicator of the market viability of the vendor and will have a significant impact on your subscription costs in the long-term. Don’t fall for claims of SaaS when all the vendor is offering is a piece of traditional web-accessible software with a pay-per-use pricing model bolted on top.
[Vamshi is the newest addition to Pi team and he will mainly cover enterprise software and cloud computing opportunities in-depth. If you are a startup/business in cloud services, do connect with us]
