Comparisons and advocacy

 
 
Picture of Peter Seaman
Centralized vs distributed models for LMS adoption
Group Particularly helpful Moodlers

Some of you probably heard that Desire2Learn experienced a massive outage last week:

http://www.insidehighered.com/news/2013/02/04/desire2learn-experiences-major-service-stoppage

D2L was unavailable to many customers from around noon on Tuesday til late Friday afternoon - over 72 hours of pretty much complete outage.

I know a lot of people affected by the outage, and I heard their frustration but more so their sense of utter helplessness.  There was nothing they could do except wait for D2L to come back up.  Even if they had backed up all of their files outside D2L (yet another good argument for making backups), there wasn't much they could do beyond making resources available from another server, collecting assignments via e-mail, and the like.

The situation made me wonder anyone using Moodle has ever suffered a massive outage like this (over 72 hours of unavailability). It also made me think about centralized and de-centralized models for LMSs.  Seems to me that Moodle must fall more into the de-centalized camp: if your instance of Moodle goes down and you have your courses backed up (and don't we all have them backed up??), you can create or find another instance of Moodle and load your courses and off you go.  The D2L failure was so debilitating precisely b/c there is no other instance of D2L (besides the one you are paying for). Methinks this might be called a case of "single-point failure," as far as LMSs are concerned.

One other thought that occurred to me centered on the issue of who's responsible when something goes wrong with an LMS. I've written previously in this space about a primary motivation for many organizations to adopt a fully hosted service model: when the system goes down, someone else is responsible for bringing it back up.  But in the case of D2L's big outage last week, the main disadvantage was ... someone else is responsible for bringing it back up!  That is, you are completely and totally dependent on someone else. I suspect there are many folks in the Moodle community who relish fixing stuff when it breaks, and it would absolutely *kill* them to have to sit on their hands waiting for a service provider to bring back service. So what is it, I wonder, about some folks in the Moodle community that gives them the confidence to know they can host their own Moodle and bring it back when it goes down?  Is it b/c they know they have the resources of the worldwide Moodle community backing them up?  Is it b/c they know Moodle is a fully tested, really robust LMS that will run just fine, meaning any service problem will be external to Moodle itself?  I'd like to know where this confidencer comes from.

One other point about LMS architecture:  One selling point of D2L for many instructors is the fully self-contained e-mail system (forwards outside D2L but e-mail is otherwise completely self-contained). The advantage is instructors can keep all course e-mail contained within the LMS, but again, that's the disadvantage when the system goes down. Here we see the wisdom of the "hub and spoke" architecture of an LMS like Moodle: don't build your own e-mail system - link to existing e-mail systems.  Likewise for content: the less that lives within the LMS, the less you lose when it goes down.  Makes me wonder whether all LMSs will one day function like a "spine" or "central nervous system," coordinating between SISs, CMSs, external e-mail systems, etc.  The LMS grabs the relevant info and resources from other systems and makes them available as needed.  Maybe it's a pipe dream, but the D2L outage shows the other side: when everything is fully contained in one system, you have *nothing* when it goes down.

In any disaster, there are always lessons to be learned, so I hope a few of you will have some wisdom to impart about how Moodle is or is not positioned to avoid a disaster like the one D2L faced last week.  Thanks.

Peter

 
Average of ratings: -
Tim at Lone Pine Koala Sanctuary
Re: Centralized vs distributed models for LMS adoption
Group DevelopersGroup Documentation writersGroup Particularly helpful Moodlers

I think there are paralleles with the question of whether it is safer to drive yourself in your car, or take a scheduled plane or tran (http://traveltips.usatoday.com/air-travel-safer-car-travel-1581.html).

Most people feel safer in a car, since they are in control. In a plane, some unseen pilot is in charge. However, the plane is better maintained than your care, and the pilot has had more training in flying then you have in driving, and the statistics back this up. Flying is safer. Then, on those rare occasions when a plane crashes, it makes the news worldwide.

 
Average of ratings: -
Picture of Peter Seaman
Re: Centralized vs distributed models for LMS adoption
Group Particularly helpful Moodlers

That's an interesting analogy, Tim. I do enjoy pulling analogies apart, so let me just see how this one does or doesn't hold up.

I think one thing that makes air travel really safe the world over is the worldwide scrutiny that exists over every aspect of air travel. The moment a rivet on some airframe fails in Tokyo or Sydney or Miami, everyone in the world knows about it and is inspecting the rivets on their own planes, authorities are issuing advisories and possibly grounding planes, etc. Because of the paramount need for safety in the aviation industry, all (or most) aspects of the industry are fully open to inspection at any time.

Seems to me that Moodle is more like this than a commercial LMS ever could be. When the code is available to everyone in the world and you have thousands of pairs of eyes looking at it, doesn't that make it more likely that experts of all stripes will spot problems and help to fix them?

Let's look at the pilot: when I select an LMS, I really know nothing about the folks who are flying that particular plane.  I know when I step onto a commercial airliner, the pilot is fully qualified, has thousands of hours of training, is sober (usually), etc.  But when we select a commercial LMS, we really know nothing about the folks building and operating it (nothing beyond reputation). In this sense we are "flying blind" when we pick a provider, since there are no standards in the LMS industry. The open-source LMS movement openly acknowledges the lack of standards and we all basically agree to do as well as we can together (a squishiness that drives accountability-minded people away, in my experience).

So I think the analogy is good but incomplete. It echoes the mentality many resource managers have in higher ed - namely that "If I leave the flying to someone else who says he knows what he's doing, I'll get there safely."  Last week's D2L outage clearly shows that is a false assumption. I just wish I knew how to make an argument for equivalent reliability of an LMS like Moodle.

Peter

 

 
Average of ratings:Useful (1)