[Replication] Terracotta - Maybe We Can Juice It?

[Ole Ersoy <ol...@gmail.com>]

http://www.theserverside.com/tt/articles/article.tss?l=TerracottaScalabilityStory

I've been peeking at this, and seems like
Terracotta might have a stuff in common with ADS.

I'm at 50K feet right now, but it would be interesting
to see what API overlap there is and if there is overlap,
which API performs better.

The reason I think there's overlap is because Terracotta is
a server that efficiently synchronizes data between multiple
JVMs.  So I think there might be something to learn from it's
network communication layer, especially with respect to synchronization
  / server replication.  I'm totally shooting from the hip though.

Cheers,
- Ole

Re: [Replication] Terracotta - Maybe We Can Juice It?

[Ole Ersoy <ol...@gmail.com>]

Just swinging from your "low level control" statement...maybe
that's something they want too...

Some of the stuff that he brags about in the article sounds impressive,
but it could be that it's only because I'm still trying to figure
out how to spell multi-master.

Cheers,
- Ole



Chris Custine wrote:
> I'm no expert on either Terracotta or Multi-master replication, but I 
> think that one of the problems with these clustering or caching 
> mechanisms is that ultimately you have don't have enough low level 
> control of the messaging between nodes which I think is extremely 
> important in developing multi-master....  Also, I believe that 
> Terracotta requires a central hub for distributing data so therefore 
> doesn't even support multimaster (this was the case last year at least, 
> not sure if it has changed).
> 
> Chris
> 
> On 5/25/07, *Ole Ersoy* <ole.ersoy@gmail.com 
> <ma...@gmail.com>> wrote:
> 
>     http://www.theserverside.com/tt/articles/article.tss?l=TerracottaScalabilityStory
> 
>     I've been peeking at this, and seems like
>     Terracotta might have a stuff in common with ADS.
> 
>     I'm at 50K feet right now, but it would be interesting
>     to see what API overlap there is and if there is overlap,
>     which API performs better.
> 
>     The reason I think there's overlap is because Terracotta is
>     a server that efficiently synchronizes data between multiple
>     JVMs.  So I think there might be something to learn from it's
>     network communication layer, especially with respect to synchronization
>       / server replication.  I'm totally shooting from the hip though.
> 
>     Cheers,
>     - Ole
> 
> 

Re: [Replication] Terracotta - Maybe We Can Juice It?

[Chris Custine <ch...@gmail.com>]

I'm no expert on either Terracotta or Multi-master replication, but I think
that one of the problems with these clustering or caching mechanisms is that
ultimately you have don't have enough low level control of the messaging
between nodes which I think is extremely important in developing
multi-master....  Also, I believe that Terracotta requires a central hub for
distributing data so therefore doesn't even support multimaster (this was
the case last year at least, not sure if it has changed).

Chris

On 5/25/07, Ole Ersoy <ol...@gmail.com> wrote:
>
>
> http://www.theserverside.com/tt/articles/article.tss?l=TerracottaScalabilityStory
>
> I've been peeking at this, and seems like
> Terracotta might have a stuff in common with ADS.
>
> I'm at 50K feet right now, but it would be interesting
> to see what API overlap there is and if there is overlap,
> which API performs better.
>
> The reason I think there's overlap is because Terracotta is
> a server that efficiently synchronizes data between multiple
> JVMs.  So I think there might be something to learn from it's
> network communication layer, especially with respect to synchronization
>   / server replication.  I'm totally shooting from the hip though.
>
> Cheers,
> - Ole
>
>