Dave Rosenberg has written a thoughtful post on Open Source vs. SaaS with a declaration “Open Source will Win“. His key argument is that open source creates a community of users and developers while SaaS only creates a community of users. I think that open source is a powerful (and rather old) distribution mechanism that has several benefits, and in that sense it is a winner.
Right Facts, Wrong Conclusions
But, Dave then implies that open source software companies are commercially more successful than SaaS by pointing to recent exits by Zimbra (acquired by Yahoo!) and XenSource (acquired by Citrix). The argument appears to be that there is somehow a choice between open source and SaaS and that SaaS will loose. I think he is wrong on this and let me take on his arguments one at a time:
- Developer Communities: You would have to be from another planet to argue that software from companies like Oracle, Microsoft, IBM, etc. do not have a strong developer community. Yes, many developers in these community work and contribute for money and recognition – and not necessarily for the ‘love of software’ but that’s a socialism vs. capitalism debate I would leave others to decide upon. Other examples of successful developer communities that do not revolve around open source include EBay, Google, etc.
- Zimbra & XenSource Acquisitions: Hotmail (probably the very first consumer SaaS company) was acquired for hundreds of millions of dollar almost a decade ago. I am not sure the Zimbra acquisition proves the value of open source – in fact, it may prove that you need a hosted runtime platform (such as Yahoo! or Google) to derive value from software. The XenSource argument is completely lost on me – just look at the valuation of proprietary VMWare. I do agree with Dave that building your business on top of one proprietary platform that is not standards-based (Salesforce) does not leave you with too many exit choices. But that is a Salesforce/ApEx issue – hundreds of companies have been acquired that built their business using components and platform technologies that were not open source.
- 99% of all SaaS runs on Open Source: Dave argues that new school SaaS companies are built on open source. This is very hard to argue – yes, a very large %-age of all SaaS (and non-SaaS) companies leverage open source. However they do not run on open source only. If you are a successful, profitable SaaS provider that needs to meet SLA’s and provide security and privacy assurances to customers – chances are you are using enterprise grade technology as part of your platform. The poster children of SaaS including Salesforce, NetSuite, Rearden Commerce, etc. all leverage software from open source and non-open source camps.
Dave further argues and I quote
The SaaS guys that provide actual value like Omniture, OpSource and a few others have a retainable space in the ecosystem. But, it’s hard to see how majority of the companies on display at Dreamforce will build sustainable revenue outside of the Salesforce.com
I agree with Dave that building your entire business to rely on one proprietary platform with its own language (ApEx) is probably not a great long-term strategy. I would have concluded that you should write your applications (SaaS or on-premise) using standards-based technology platforms (like Oracle’s) and not bind to a restricted proprietary language/platform.
SaaS Providers Profit from low TCO platform (including open source)
The SaaS Providers (as any other software enabled business) look to keep their costs low and this requires careful evaluation and adoption of many of the following:
- Low cost hardware: This may mean commodity AMD/Intel boxes or, as Jonathon Schwarz would argue – it could mean leveraging high-end Sun servers to keep total cost down including cost of power, cooling, operational management etc.
- Scalable/manageable/customizable software: In order to scale with growing demand (and isn’t that what every successful business hopes for) – you must have software that is easy and cost-effective to manage and support. Linux is an excellent example of open source software that comes with support from likes of Oracle (my employer, see full disclaimer below). The software stack could also include a cost effective database or a rich customizable business process tier (middleware). The issue is not access to the source code but adherence to standards, tools for manageability, ability to scale, support for security, etc.
To decide your platform on a single dimension (access to source code) would be like purchasing a car because I have access to detailed designs. It could be the absolute right criteria for a hobbyist that wants to re-do the engine but for most buyers the question is much broader and requires holistic thinking about one’s objectives.
Hat tip: I came upon Dave’s post via Matt Assay (CNet Blog).
Update: Treb Ryan carries the conversation further to a conclusion on his blog. But he raises an interesting point about how SaaS and open source don’t really play well together apart from the obvious use of some open source technologies by some web applications. I think this will lead to more blogging, so its unlikely to be the Final Word!