13.1 Running a service is not a system  (Page 6/7)

Thanks, Ken

4. dick moore says - july 30th, 2007 at 3:33 pm

Lots of issues there Ken

I would first like to tease out a little more of your thinking around the connection between supporting individual users that expect a unique experience and using open source software

Hmmm while there is no direct connection, using open source software has allowed us to build highly resilient infrastructures that scale. We have a finance and Management Information system that provides performance data a week in arrears.

Using OSS and commodity hardware has meant that that the system has had 100% uptime over the last two years and we are able to scale horizontally keeping pace with demand (downloads have never taken more than 10 seconds)

Here is my first assertion. It seems that the “customisation” criteria in the above mix is most critical, after all, high volume and high reliability are pretty typical reasons to out source.

It’s not the customised end user experience that makes in-sourcing important, it’s the degree to which we wish to customise the end user solution that is the driver. If your solution is subject to minor change then our-sourcing is a good option. Our e-learning platform is our business and we want to be able to make significant changes on a regular basis. Trying to contract manage third parties to both provide 99.9% uptime AND process lots of change is not easy, contracts by their nature are all about defining risk in advance. In-sourcing has allowed us a finer grained management of risk.

Through your experience, what advantages does OSS potentially provide that proprietary options do not?

In many respects using OSS has a similar risk profile to in-sourcing. As a purchaser you always take the risk. Using commercial software you are buying into a solution with the intent of reducing risk. Of course this is often, (though not always), a chimera, commercial software comes with a service contracts and SLA’s though when one hits an significant incompatibility, it’s either very expensive or impossible to have it customised for your application. Don’t misunderstand me, we use commercial and OSS database software. For all critical data I use the commercial provider. Our open source database software provides fantastic value for those applications that require read-only access.

And, when you are evaluating OSS options, what are some of the evaluation criteria that you prioritise?

Good question, we would use much the same criteria that we would for commercial software.

• How well established is it
• Is it a market leader
• What is the size of the user base
• Are there third parties providing support
• What do the technical people I respect think of this product
• How good a fit is it with our other Technical Roadmaps
• What is the total cost of ownership over 3 years likely to be

Using the above criteria to evaluate Apache, against other web servers , we decided to use apache :-)

Hope that helps Dick

5. ken udas - august 2nd, 2007 at 7:27 am

Dick,

Thanks much for this. I am sort of trying to make some connections between what we have learned through your posting and previous postings. Customization/localization is a major theme in both the open source software postings and the open educational resources postings in this Series, which I find quite interesting. You have introduced a different (or at least what feels like a different) aspect of customization. To this point it seems that most of the dialog about customization has been in recognition that different groups of learners (and faculty and administrators) will have different needs, so content and infrastructure should be localized to meet local cultural, linguistic, access, etc. circumstances and needs. It has been indicated generally that OSS and OER provide better opportunities for localization than proprietary software and educational content. Mara, in our last post, also pointed out some of the challenges associated with the level of customization that OSS can provide can impact on usability testing and user experience.