SaaS and Cloud Computing - Mainframes come full-circle
by Sarat Pediredla Under General, RantsI was on the panel last week, at a very interesting Think and a Drink event on SaaS and Cloud Computing.
For the purpose of this blog post (and given the gist of the event), let us simplify "cloud computing" to mean "utility computing" (although the purists would argue the cloud is more than just this), where the idea is that you can access, use and discard computing power and resources in the same way you get your gas or electricity, i.e., pay as you use. Also bear with me while I throw in random musings about SaaS in the same breath as cloud computing, since they are so inter-linked.
The core debate of the panel was on the pros and cons of cloud computing, with some having a bi-partisan view on whether it is good or bad for you. However, the undercurrent was on how this is the hot-topic of the day and how everyone could utilise this "revolutionary" advance in technology. Certainly, the recent marketing overdrive by giants like Google, Amazon, and IBM would have you believe that (Random alphabet here)aaS is the future of computing.
Now, I am not nearly as old to say this with any authority, but wasn't this exactly what mainframes offered in the old days? Sure, the Internet makes this cloud a lot more accessible, and huge compared to primitive mainframe deployments, but the concepts of time-sharing, using processor cycles and dumbed down terminals are certainly not new. In fact, the whole desktop/personal computer market came to be because people were dis-satisfied with the restrictive nature of using software that was under a centralised regime.
The real challenge for people in decision-making positions in terms of organisational infrastructure is not to take the extreme viewpoint but to judge the necessity for cloud computing and it's use on a case-by-case basis. In summary, I think the following key points were made by the panel,
- There was overall consensus that "cloud computing" was too often being used as buzzword to cover a lot of scenarios and that we have all been using SaaS from the days of Hotmail and Salesforce, well before Web 2.0 made it trendy.
- SaaS and Cloud Computing deliver real value to younger, smaller and riskier start-ups by allowing them to create and run an infrastructure that can compete with larger competitors. It was agreed that larger enterprises would rarely see the benefits of increased adoption.
- The main arguments for SaaS (especially) was that there is a lower barrier to entry and you don't need an "IT Department" to deploy and run software that is critical to your business. I personally see this as being a problem with traditional "deployed" software, with it's lack of user friendliness, rather than a real advantage of SaaS.
Salesforce became the dominant player in the CRM market not because it was hosted, but because it offered a painless and user-friendly alternative to the consultingware offered by competing offerings like Siebel and SAP. - There was also a general consensus that SaaS and cloud computing generally cost more in the longer run due to the popular subscription model and the need for people providing hosted services to earn recurring revenue.
- One argument I do disagree with in the whole hosted vs installed debate, is that systems you own and run are somehow inherently more secure and reliable. The biggest myth when it comes to the SaaS debate in the enterprise is the concern over data security in hosted systems. Given that SaaS providers cater to economies of scale, their reliability and data security will more likely be better than most in-house infrastructure can come up with.
- Flexibility is a huge problem when it comes to hosted software. Again, economies of scale dictate that software delivered using the SaaS model is mostly homogenous, which means you have little or limited choice when it comes to functionality (Facebook re-design anyone?)
- A great practical application of SaaS and Cloud Computing infrastructure is to outsource non-core and non-critical operations to these systems. For example, at hedgehog lab, we make heavy use of the cloud to deploy bandwidth intensive downloads (Amazon S3) and we are experimenting with using Amazon EC2 to provide us with a wealth of virtual machines to run our integration and test environments.
However, it was very clear that a lot of enterprises still have real concerns about basing their business model and/or their core operations on a SaaS or cloud computing provider.
I have no doubt that as hardware becomes cheaper and skilled resources become dearer, more people will opt to move everything from non-critical to core operations into the cloud, and that SaaS adoption will increase gradually.
However, let us not lose sight of the reality of the software industry as it stands. Desktop and installed software still outsells hosted software (I would love to know the stats if anyone has them) and SaaS/Cloud Computing is no silver bullet.