river-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "John McClain - Sun Microsystems, Inc." <John.McCl...@Sun.COM>
Subject Re: a potential "AR1" release
Date Fri, 31 Aug 2007 13:18:50 GMT
Mike Morris wrote:
> John McClain - Sun Microsystems, Inc. wrote:
> 
>> My vote is to remove logstore altogether and do it for AR1.
> 
> +1 -- good candidate for re-inclusion in a later release.

Probably should have been a bit more explicit before, but at the time I 
didn't have quite enough energy to write a longer message....

In my mind this would be a one way trip for logstore, it wouldn't be be 
coming back.

The problem with logstore is that it depends on Progress Software's 
ObjectStore PSE Pro. At build time you need a post processor that comes 
with PSE Pro (though there might be alternatives here), and at runtime 
you need their library, pro.zip. Getting access to either requires a 
license from  Progress Software, which in turn generally requires $$$.
Because of this I don't see having a dependency on PSE Pro was being an 
option for River.

Since 2.0, Outrigger's store has been plugable, and there is a second 
store implementation in the starter kit, "snapstore". Snapstore does not 
depend on any external libraries. Snapstore doesn't scale as well as 
logstore as the number of entries in the space goes up, but storing lots 
and lots of entries has never been Outrigger's design center, so in 
context this may not be a big limitation.

It is worth noting that because Outrigger caches the entire data set in 
memory (hence the note that "storing lots and lots of entries has never 
been Outrigger's design center" above) writing a new store isn't that 
hard.  If snapstore turns out not to be acceptable I can think at least 
of 3 other options for new stores that we could pursue. There are also 
some improvements we should make to snapstore that while they wouldn't 
make it scale better in the total number of entries, they would 
substantially improve its multi-threaded performance and cut the latency 
of commits dramatically.

Don't know if this changes anyones mind.

-- 
John McClain					john.mcclain@sun.com
Sun Microsystems, Inc.
Burlington, MA

And it is that way today. We are tricked by hope into starting
companies, beginning books, immigrating to this country and investing
in telecom networks. The challenges turn out to be tougher than we
imagined. Our excessive optimism is exposed. New skills are demanded.
But nothing important was ever begun in a prudential frame of mind.

         - David Brooks



Mime
View raw message