Fault Tolerance?

Matt Ingenthron Matt.Ingenthron at Sun.COM
Fri Sep 28 16:54:18 UTC 2007


Dustin Sallings wrote:
>
> On Sep 27, 2007, at 21:56, Paul Scott wrote:
>
>>> Agreed. Wouldn’t it be great though to have a mem-based HA datastore?
>>
>> I would certainly vote +1 on that idea!
>
>     You do realize you wouldn't have anything remotely like the 
> performance of memcached, don't you?  You'd need something along the 
> lines of two-phase-commit if you want any kind of correctness.  If you 
> don't want correctness, then why are you worried about HA?
(snip...)

I don't know if it's useful, but it seems on topic....

Sun recently built, entirely in Open Source (both CDDL and GPLv2), an HA 
object data store in the Shoal project[1].  In fact, the docs lead one 
to believe it's just an interface that would need to be implemented 
(though I suspect it may be slightly more complex than that) [2].  At 
one point in time, they had on their list of thoughts to add on memcache 
protocol.  This would be useful for anyone using existing libraries in 
PHP, RoR, etc. who needs a highly available session/state object store.  
In fact, shoal was designed for exactly this in Glassfish v2: it's the 
primary HA mechanism.

I agree with Dustin, et. al. that we'd be talking about something very 
different than memcache though.  Doing this correctly would be a large 
departure from the high-performance simplicity of existing memcached.

That doesn't negate the fact that it would be useful to have something 
that speaks memcache protocol that handles a lot of the complicated 
aspects so you don't have to rewrite portions of your app and/or use 
different libraries.

- Matt

[1] https://shoal.dev.java.net/
[2] https://shoal.dev.java.net/ShoalOverview.html


-- 
Matt Ingenthron - Web Infrastructure Solutions Architect
Sun Microsystems, Inc. - Global Systems Practice
http://blogs.sun.com/mingenthron/
email: matt.ingenthron at sun.com             Phone: 310-242-6439



More information about the memcached mailing list