invalidation by namespace Re: Listing all keys on a server: why?

[Chris Hondl]

We've implemented essentially this approach to namespaces in PHP here at
IMVU (we call them a cache-class).  It works great - we set very long
timeouts on most of our queries, and still give our users instant feedback
when the underlying data changes.

I've blogged about our implementation
here<http://www.imvu.com/blogs/index.php?blog=793&title=extensions_to_php_memcached_api>and
included a zip file containing our implementation for others to use.

Chris

-- 
http://www.imvu.com/technology

On 7/28/06, Serhat Sakarya <serhat at sakarya.nl> wrote:
>
> On 7/28/06, Paul T <pault12345 at yahoo.com> wrote:
>
> >
> > 3. key_salt is stored in memcached forever.
>
>
> The problem I see with this approach is that when the salt *does* get lost
> (node reboot, failure, etc), you risk that old data may be revalidated.
>
> In this case, I think that it's better to trade some efficiency for data
> integrity by including the salt in the value instead of the key. It will
> result in more lookups, but losing the salt won't cause issues. The
> scalability in this area makes sure that any performance loss won't be of
> the painful kind.
>
> Regards,
>
> Serhat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.danga.com/pipermail/memcached/attachments/20060728/88e76926/attachment.html