[DG: Open Forum] Sakai OAE

[David Adams]

Mark Norton wrote:
> How much change to the architecture will be needed?  Is it "back to the
> drawing board" or just a matter of tuning things up?

After four years, the system can't support more than a few dozen users
at a time, or search through more than a few thousand objects.
Meanwhile, the potential customers need to operate and hundreds or
thousands of times that scale. The architecture has been flawed from
the start; built on false assumptions, buzzwords, and crossed fingers.

In designing the OAE, all the clever and interesting parts of the
CLE's design were thrown away, and all the weird and broken things
about the CLE's design were doubled-down on. Anti-RDBMS sentiment has
been behind most of the poor design and performance problems in CLE,
and we see that things are that much worse now that the RDBMS has been
eschewed entirely. The ironic part is, the NoSQL storage system itself
is actually just an application that uses an RDBMS for storage.

Developers hate thinking about how the data gets stored and having to
worry about optimizations, but adding another layer of abstraction
does not eliminate those details, it just makes them much harder to
address. RDBMSes may be boring and awkward, but they're mature, well
understood, well supported, and easy to optimize. Yes, it's old
technology, but the reason we still have relational databases is that
they actually turn out to be an amazingly efficient way to store data.

>>  OAE does not have all the features of an LMS
>
> How big a gap is this?  Is there a chart of targeted features with
> status?  If not, why not?

The gap is complete. OAE has no distinctly LMS-type features. Vague
nods to the need for these have shown up in various roadmaps
(generally released each summer), but always at the elusive "stage
three" milestone that never actually gets reached. If anything, the
story of OAE is that it long ago moved from being an attempt to write
an LMS to being an attempt to write a *platform* for someone else to
write an LMS. That it's attempting to build a generic platform on top
of other generic platforms (Sling, Jackrabbit, etc) may be part of the
problem.

>>  1. and 2. are taking too long
>
> It wasn't a resource issue until most of the big players left. Why is it
> taking too long?

If you've read Frederick Brooks's The Mythical Man-Month, it's no
mystery. The lack of technical leadership, poor communication, lack of
documentation, vague design goals, the Second-System Effect,
inconsistent staffing, mission creep. As to our current situation,
facing a broken system that needs a redesign, we should turn to John
Gall's The Systems Bible: "A complex system designed from scratch
never works and cannot be made to work. You have to start over,
beginning with a working simple system." To write a successful complex
LMS, we would need to start from a successful simple LMS.

The Sakai Foundation should take this as a perfect opportunity to
abandon the OAE project and direct whatever resources are left into
improving the CLE. For years now, the CLE has stagnated while the OAE
was sold as being just around the corner, highly capable, and
providing a smooth transition from the CLE, none of which was true.
It's time to set the record straight, admit the OAE has failed, and
save whatever credibility is left. Put the resources that still exist
into improving the working system we do have instead of pouring
endless resources into a non-working system that has no chance of ever
being fixed.

-dave