Forum OpenACS Q&A: Professional OpenACS Setup/Op Costs

Request notifications

Posted by Bob OConnor on

Here is an opportunity for a shamless plug by three companies that offer "professional help".

"Bob, YOU need professional help!" ;-)

The cool new OpenACS site home page mentions: Furfly, OpenForce, or Ybos as three companies that offer services.

I would be interested in:

  • what are the costs to set up and operate an OpenACS site?
  • what are the hourly costs for help in OpenACS development and operation?
  • What can a single developer expect to be paid to set up and operate a site?
  • Ok I know the complexity of a site is a factor... What about using a "typical" site, if there is such a beast, in order to get a ballpark idea. I understand that ArsDigita charges somewhere around $20,000 per month and this seems high for any group starting out small unless they get vulture capital to get started and continue to operate.


    If you don't want to reply publically, you can send me a private email.

Posted by Lee McKusick on
Well, it has taken me a calendar month, maybe 80 working hours to set up a copy of OpenACS 3.2.

I expect to expend another 200 hours labor to learn TCL, learn PostGres, configure a chroot jail,  and grok the data structure.

Direct costs associated with my set up are approx $900 for server, $100 for a UPS, $50 for Philip and Alex's Web Guide book.

Costs not incurred yet are the charges for a high bandwidth internet connection.

Another cost not expressed here is the hours to find sponsors and clients for my particular non-profit human service vision.

So, using Lee's Law of Three-to-Five [ Any really new software project really costs Three to Five times as much as you thought It would ] and $25 /hr for labor...

An OpenACS installation with any significant measure of customization is easy to price out at $8500 (during the honeymoon phase of the project) and actually cost $25,500 to $42,500.

Most of the cost is "labor" and clearly a learning curve must exist and the learning curve must be very steep, because this is software and there are  many discrete systems within the OpenACS concept.

Personally I am a wussy about this stuff and I'd  love a series of RPM files to make setting up OpenACS a snap, and bye and bye maybe that will happen.

And I post this estimate with affection and apologies to the wonderful people working to make OpenACS happen. I could be wrong by a factor of 3x-to-5x either up or down. Clearly, if I had more experience in TCL, Postgres, SQL or AOLServer the hours would be less.

Posted by Ben Adida on
Lee, my question is, now that you've done it once, how much do you think it would take for you to set up a new OpenACS installation? The first install is going to be painful, especially if you've never seen AOLserver, Postgres, and ACS. However, the second one should be a different story.
Posted by Li-fan Chen on
That's an exciting point Ben pointed out. Imagine if those Mac GUI guys working on Eazel and remote system administration (I believe the correct buzzword is Zero Administration) could cook up an wizard or remote administration database around OpenACS administration and setup tasks! :-) Sounds like the logical second step once we start getting used to RPMs and DEBs and get our own CPAN for ACS.
I wonder what it will take. They propose to directly manipulate system dot and etc files (being able to parse and change them as well as a real human admin I guess) and access the greater desktop via GNOME's BONOBO/CORBA/(gnome idea of a registry) facility... so maybe someday.. far in the future.. we could provide a BONOBO api to ACS/PostgreSQL/AOLserver/ACS modules settings file so the Eazel guys can tinker with it from afar. It would be nice if the ACS community can do the tinkering and provide the remote administration though, so I hope that remote administration facility Eazel will be selling by the seat will also be open-sourced. I guess someone should go and ask them what the deal is.
Posted by Li-fan Chen on
Lee has pretty much set out a clear balance sheet for all to see though, so I have a few comments about the items he listed as taking his time:
  • 1) When Lee says spending time to learn Tcl, does this include reading quite a bit of all the ADP and Tcl files that comes with ACS, or are we talking about just learning Tcl?

    I sort of assume most anyone in the ACS community did just this if they have created a few sites integrating ACS. Or if they have come up with ACS modules extensions or even done major ports and enhancements (Don and Ben and friends) There's a big difference between reading code examples in Brent's book and reading a quarter of a million lines of Tcl in ACS.

  • 2) Since Lee mentions the data structure of the ACS, I like to comment that the data structure can be a bit of a puzzle -- at it is to a SQL baby like me. I think that means you end up having to read a lot of source (both SQL and Tcl and maybe C) before doing great things with the kit.

    However, the nice thing about going overboard to learn the entire system line by line is that changes made to the distribution you download fro or won't come as too big a surprise--and will be easier to cope with (*duh!*). As an added bonus, you have the potential opportunity to chip your big ideas to Ben/Don/Arsdigita/friends. Or at least know how to create a well rounded ACS-aware module instead of a plain ol AOLserver Tcl script package. Finally, if you make webpages with AOLserver, you'll also learn how to express yourself in Tcl/SQL by imitation.

  • 3) Lee's "balance sheet of developer time" makes no mention of what I call paranoid over-preparation (which is quite costly and a luxury few development firms can afford to enjoy...umm... actually.. there's HP: they have a tradition of promotion of enterprising employees--some turning into company-paid scientists!).

    ArsDigita and OpenACS are represented by ol' farts (okay maybe not so old *wink*) who really know their stuff (going all the way back to VAX assembly and LISP machines, I'm afraid :-)--whatever their background. They all have timeless scary war stories to share and reuse before their arrival on the Tcl, ADP or SQL scene.

    Point is, as a part of the younger generation of programmers (that includes me) we have to go the extra mile to absorb all sorts of technical know hows besides what's immediately related to the ACS before feeling like we are seeing what others see when we work with the ACS system (and quite unfortunately, as Lee illustrated quite optimistically: ACS is a oceanlinerful of technologies all by itself without even taking the scenic route).

    Frequently are times when I feel terribly inadequate coping with even the high level buttons and knobs of ACS :-) I usually cope with this lack of experience by trying to do some unique things with ACS itself or study and write lots of related code (in any language) from projects that vaguely promotes the same dreams ACSers aim for. Coincidentally, the current super nova of confidence in open source development, simplicity of Tcl and the ubiquitous web infrastructure greatly simplifies any newbie's task in their aim to learn and expand their horizons. Combining the ACS kit with un-heard of features, applications or technologies is now straightforward and there's no better time to start.

  • 4) Anyway, it can be a fun experience for many (even eventual OpenACS customers who comes here looking for someone to fill contract!) to get caught up in this evolutionary idea of web community building. Hopefully the time or money spent on the kit will ultimately magnify the ability for netizens to communicate with one another and do commerce like never before.

    From the developer's point of view, if you can make money doing something socially positive with ACS, that makes it a big bonus over the competition.

  • **time to turn the super-charged bull shit rant mode off**