jakarta-jcs-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Aaron Smuts <aasm...@wisc.edu>
Subject RE: diff between Remote Cache and Lateral TCP Cache
Date Sun, 20 Jun 2004 18:00:26 GMT
Are you using the remote cache or the lateral?  

The remote cache should suit your needs.  I would never use the lateral
for gets, except to cluster remote caches, maybe.

The javagroups auxiliary would meet your needs and so will the remote.  

The remote cache can be configured to issue remove requests when an item
is put or to broadcast the item.  Actually, it is a client
configuration.  The remote cache just sends the object and the client
gets to decide what to do with it.

Aaron

> -----Original Message-----
> From: Matthew Cooke [mailto:mpcooke3@lineone.net]
> Sent: Saturday, June 19, 2004 1:52 AM
> To: Turbine JCS Users List
> Subject: Re: diff between Remote Cache and Lateral TCP Cache
> 
> I would guess (and this is just a guess) that most people would prefer
a
> faster JCS that you could reliably pull out contents from in
millisecond
> time than a fully distributed index.
> 
> The reason for this is because for most people rapid get() speed and
cache
> consistency are probably more important than scalability of the level
> currently being described on this list. I am thinking for instance of
> people requiring database or HTML caching in a medium sized Java
webapp.
> 
> We only have a 12 machine cluster and already remote retrieval on
get() is
> causing some problems and means we can't use it to cache large objects
due
> to the retrieval time.
> 
> It's not that I can't see uses for a fully distributed index, in a
much
> larger system where cache consistency isn't important like google's
> caching
> system.
> 
> Is there any configuration of JCS that performs fully distributed
remote
> put's  whenever any cache recieves a put() request? Is there anything
that
> provides reliable IP multicast based push, what about the javagroups
based
> auxillary?
> 
> Matt.
> 
> Travis Savo wrote:
> > The objective is to have a distributed index. If you put locally
from a
> > lateral get, your duplicating the index. Lateraling allows you to
have
> an
> > index that's divided among the caches, yet still appears as one big
> index to
> > any given client. Whoever gets the object first is the owner of the
> index,
> > which is (in theory) not in any other cache, yet still accessible by
all
> > caches.
> >
> > Weather or not this is a Good Thing(tm) depends on your usage
> > characteristics.
> >
> > -Travis Savo <tsavo@ifilm.com>
> >
> > -----Original Message-----
> > From: Hanson Char [mailto:hanson_char@yahoo.com]
> > Sent: Friday, June 18, 2004 3:41 PM
> > To: 'Turbine JCS Users List'
> > Subject: RE: diff between Remote Cache and Lateral TCP Cache
> >
> >
> >
> >
> >>Lateral cache treats all caches as peers. When a cache gets a cache
> miss,
> >
> > it asks all the other caches if they have the object in question.
> >
> >>If one of them responds, the object is returned as a cache hit, but
IS
> NOT
> >
> > stored locally
> >
> > Out of curiosity, why is an object obtained from the lateral peers
not
> > stored locally (which would speed up subsequent local access) ?  I
mean
> > what's the rationale behind this design ?
> >
> > Hanson
> >
> >
> >
---------------------------------------------------------------------
> > To unsubscribe, e-mail:
turbine-jcs-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: turbine-jcs-user-
> help@jakarta.apache.org
> >
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
turbine-jcs-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail:
turbine-jcs-user-help@jakarta.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: turbine-jcs-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: turbine-jcs-user-help@jakarta.apache.org


Mime
View raw message