Internet has had a profound impact on the way we interact and work. It started of as a medium to exchange information between computers. With increasing network speeds and wider penetration of Internet, software providers started moving their applications to the Web, and started offering their software’s as a service (SaaS). Organizations or individuals could now use enterprise level software’s for a small fee (or even for free) without having to install them on local infrastructure, and without worrying about how to maintain these applications.
But all this while there was almost no innovation on how hardware infrastructure was deployed or licensed. One either had to create an in-house data center to deploy and manage their software systems, or had to take servers on rent in a data-center managed by someone. In both cases one had to pay full server and bandwidth charges even if their applications were consuming only a fraction of these resources, or running for only few hours in a month.
Cloud Computing or Infrastructure as a Service (IaaS) changes all this. Using cloud infrastructure provided by service providers, such as Amazon or Microsoft Azure, you can now –
- Choose the server configuration for your needs
- Activate/deactivate it instantly and without any charges
- Increase/decrease number of server as per requirement – again without any charges
- And best of all, pay for only what you consume (e.g. pay only for bandwidth that you consumed and not for some flat GB’s of data transfer, or pay only for the duration when your cloud infrastructure was ON and not for the whole month etc.)
Difference between self-managed server infrastructure, servers rented from data center management companies, and cloud based infrastructure can be best explained using the taxi analogy I saw in rSmart’s video on cloud computing –
So a dedicated server infrastructure is like having a self-owned car, which is best option if you wish to have customized infrastructure, you know that you are going to use the infrastructure extensively, or your security regulations demand the servers to be situated in your premises or within your strict control. On the other hand, leasing infrastructure in a data center is like having a leased car, where someone else is managing the infrastructure for you, and you pay a fixed fees for your servers irrespective of it’s usage.It’s a good option if you want to avoid big capex, and want to have flexibility to change service providers or servers in case you are not satisfied with the performance. Cloud computing is like taking a taxi ride, pay for what you use and how much you use, and dynamically increase compute capacity by requisitioning more servers from the cloud service providers.
There is off-course perceived sense of loss of control and security when you are not hosting dedicated infrastructure at your own end, and the costs can also shoot quite a bit in cloud compute model if your solutions are bandwidth and/or I/O hungry. So one should be careful of these trade-offs when moving to cloud computing model.
And what can cloud computing mean for e-learning industry?
Learning management systems (LMS’s) are already made available in SaaS model by many LMS providers (including us). But for most LMS providers, cost structure for providing LMS in SaaS model is currently governed by ‘old’ way of server infrastructure management (in-house or rented), which is then as-is passed on to the customer.
With cloud computing option, LMS providers will be able to create different and more flexible licensing options, such as –
- Pay $x per user per course (of a certain size) irrespective of duration for which user is registered on the system
- Pay $y per user per assessment, where assessment duration is say 3 hours, and where user may come to the system at any point of time
- Pay $z for a surge in concurrent user load from say 100 to 500 for a short duration which may be because of a timed learning event
Essentially, true ‘pay-as-you-go’ model where customer is transparently charged for the infrastructure resources consumed by their users, instead of time linked flat fee for expected registered or concurrent user load.