Regarding Consistent Hashing....

Dustin Sallings dustin at spy.net
Thu Aug 30 23:10:29 UTC 2007


On Aug 30, 2007, at 15:54 , dormando wrote:

> (Just tossing out an idea, I've already discredited myself as  
> useless here!)
>
> Would a "minimal recommended feature set" be along the lines of  
> acceptable?

	I think that's really what we're talking about.  We're just trying  
stuff to figure out what that is.

> I can imagine not having at least a common default to be insanity  
> inducing. IMHO it's probably okay to request clients at least  
> implement the crc32 over such and such details as long as they  
> default to them. Then you're still free to implement something with  
> higher performance characteristics.
>
> On your last bit ... uhhhh, no strong opinion. I think the idea  
> here is to do "something" which makes people feel comfortable to  
> actually *use* consistent hashing in some form. Presently larger  
> sites get bitten by the "I just added another memcached instance  
> and my shit got all slow!" more and more often. Lets help spare  
> sysadmins from having to do memcached maintenance at 4am? :)

	Is anyone who's uncomfortable with consistent hashing actually using  
it for the same keys from multiple different client implementations?   
If so, wouldn't it be valuable to standardize on flags and data  
encoding or something first (this was briefly mentioned in a meeting)?

	I use flags to indicate encoding mechanisms in my default  
transcoder.  Some of the bits indicate data type, and some of the  
bits indicate special processing features (e.g. gzip).

	For anyone who is accessing data from different languages, what is  
the intermediate format?  What kinds of things are you doing with it?

-- 
Dustin Sallings




More information about the memcached mailing list