Forum OpenACS Q&A: "OpenACS saved me $1 million..."

Date: Mon, 24 Sep 2001 14:41:13 -0400
To: wwwac@lists.wwwac.org
From: claygordon@mindspring.com
Subject: re: [wwwac] broadvision???
Message-ID:<Springmail.105.1001356873.0.57005600@www.springmail.com
>

> I'd be interested to hear opinions of broadvision. The
> company I work for is about to develop some sites that
> use it, and I'd like to know the pros and cons.

This comes from an evaluation of the technology in comparison with 
Vignette StoryServer, and OpenACS, the Open Source version of the 
ArsDigita Community System (ACS) (www.openacs.org). I have never 
worked with a live implementation of Broadvision (for good reason as 
it turns out, after our evaluation). Eventually, we opted for 
OpenACS.CMS can mean three things: Content Management System Catalog 
Management System Customer Management SystemIn the end, what we wanted 
to do was to serve customers, so putting the customer first, in the 
sense of ensuring that the core data structures were about the 
customer, made sense. From that core, it was  easier, we found, to 
offer all of the customization and personalization services that we 
wanted to offer our customers; it also made CRM operations to 
understand our customers easier.W!
!
hen the "C" stands for content, as it does with StoryServer, all of 
the core data structures are about how content will be moved through 
an editorial process and show up on the screen of visitors. As a 
consequence, customization and personalization are not central tasks, 
and require kludges like plugging in Net Perceptions on the side. Note 
that this is not a battle about programming languages, both OpenACS 
and StoryServer use Tcl -- however, OpenACS uses Tcl a lot more 
intelligently.With Broadvision, the catalog is the central data 
structure, and everything that the system is designed to do natively 
is about managing the catalog and merchandising (upsell, cross sell, 
etc.) to users. If you have content needs that are not merchandising, 
or customization/personalization needs that require content that is 
not in the catalog, the entire process is a real kludge.

In the end, we realized that, using BV, we'd end up spending an 
enormous amount of time and money customizing, despite the fact that 
the core task (managing a catalog, albeit an unusual one -- which is 
the problem) was something that BV was supposed to do well. The 
software demoes pretty well, but it's not that easy to use in actual 
production situations. Plus, it's slow (comparatively) and we found 
we'd need a lot more hardware for a given load than many other 
solution.We are developing an OpenACS application that uses one set of 
data structures for the editorial (back end systems) and a 
denormalized subset of those data structures optimized for delivering 
content to visitors. There is a set schedule for updating the 
production servers from the editorial servers that is based on what 
kind of information changes and its relative urgency. The servers that 
service visitors assemble pages from a database and track visitors' 
usage through a near-real-time clickstream capture and data mining 
reporting interface.There are a lot of nice things about this 
approach. The first nice thing is that we saved several hundred 
thousands of dollars in software licensing costs in the first year 
alone. OpenACS, the PostgreSQL database, and AOLServer are all 
free.The second nice thing is that we saved several hundred thousand 
dollars in hardware costs for the public server infrastructure. One 
vendor created a system that included 5 huge Cisco cache servers to 
handle the load -- based on the difficulty of an application-server 
centric model to scale to handle the load we anticipate because of the 
inherent performance inefficiencies of the level of abstraction app 
servers inherit.The third nice thing is a result of the second nice 
thing, which is that because our hardware requirements are now much 
smaller and we can serve the site out of generic servers and not 
specialized hardware, our monthly managed hosting costs are less (this 
is the cost of rack space, etc., not including the cost of 
hardware).Overall, I've documented savings of over $1 million in the 
first year, which is more than enough reason (for us) to stay away 
from products like Story Server and Broadvision, and go with 
OpenACS.Fortunately, we were lucky to find a very experienced group of 
developers.
YMMV.
Clay

I'm sure some would like business specifics; others would like technical specifics -- whatever you are able to share would be *great*!

Thank you,

Louis