Re: Dude, where´s my databa se?

On 02/02/2011 06:10 PM, Robert Friberg wrote:
>
> [...] I truly believe the prevalentsystem concept can have a major 
> impact on software development and operations.Surely, some substantial 
> percentage of projects using relational databases would benefitfrom a 
> prevalent system. You folks probably don't need convincing.
>
> Do you seasoned users agree or disagree? What percentage market share 
> do you think is viable? Why aren't we witnessing a greater popularity?
>

Excellent question! It's something I've thought about a fair bit.

As somebody who actually put Prevayler into (low-volume) production in 
2005, used a similar approach in 2007 for what became a large web site,  
and used Prevayler itself again this week for a prototype, I'd say these 
are the reasons that would keep me from using it again in production for 
anything substantial:

   1. No trusted way to handle failover.
   2. Scaling past one box (either in load or in data volume) is a giant
      cliff of work.
   3. Startup times are a major problem at scale.
   4. Upgrades require downtime.
   5. It's in Java


With regard to that last point, it's partly about how much extra junk 
code Java makes me write, and partly about the fact that Java seems to 
be becoming more and more an enterprise ghetto. Perhaps .NET is better 
for this. The tools for migrations are also rough, which wouldn't stop 
me, but it does discourage me.


And I think these are things stopping others:

   1. It's weird, and therefore scary.
   2. They're used to magic pixies (and/or surly DBAs) take care of a
      lot of stuff, like indexing, storage, and searching. Prevayler
      makes them think. E.g., command objects, baptism problems, the
      proper way to connect your objects. They don't want to think.
   3. There's no good story for how to scale past one box. With an SQL
      app, the story is "add more boxes" and/or "use a bigger database"
      and/or "give vendor X lots of money". This story is frequently
      bullshit, but it's commonly accepted bullshit.
   4. There's no good story for integration and/or reporting. Again, the
      SQL story is dubious, that doesn't stop people.
   5. It requires better developers. A lot of shops, frankly, get by on
      half-assed work. Prevayler is fantastic, but unforgiving and
      demanding. I wish that every developer had a firm grasp on CS
      fundamentals, and the ones that I hire do, but that's not true
      everywhere.


However, this is all particular to my view as a web startup guy. If I 
were building something small or isolated (e.g., internal system, 
desktop app) that I was pretty sure would stay small, I'd definitely be 
tempted to use Prevayler.


I suspect the way to get it accepted is to package it with a bunch of 
helpful tools to remove the pain points, and then use it as part of 
something awesome and compelling for developers. I don't know what that 
would be, but Rails is a good example of the phenomenon, as was LDAP 
back in the day. Ideally it should be something that makes use of 
Prevayler's incredible speed.

The only thing I can think of is the push for real-time interaction on 
the web. E.g., Quora has a fantastic notification system, and it seems 
pretty much instant. Maybe bundling Prevayler with a COMET server and 
some neat publish/subscribe stuff with a good JS component would do the 
trick.

Anyhow, I hope you take the revolution forward! It'd be great to have 
people stop saying, "Huh?" when I mention Prevayler.

William

------------------------------------------------------------------------------
Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)!
Finally, a world-class log management solution at an even better price-free!
Download using promo code Free_Logger_4_Dev2Dev. Offer expires 
February 28th, so secure your free ArcSight Logger TODAY! 
http://p.sf.net/sfu/arcsight-sfd2d

_______________________________________________
To unsubscribe go to the end of this page: http://lists.sourceforge.net/lists/listinfo/prevayler-discussion
_______________________________________________
"Databases in Memoriam" -- http://www.prevayler.org