couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tom Sante <tom.sa...@gmail.com>
Subject Re: couchdb for genome data
Date Thu, 04 Mar 2010 14:43:36 GMT
On 4/03/10 15:35, eric casteleijn wrote:
> On 03/04/2010 06:23 AM, Metin Akat wrote:
>> No, there is no limit.
>> And you can structure them in subdirectories by using a slash in the
>> database name.
>> Like "your/name/here" will create 2 subdirectories with a database
>> called "here" at the bottom.
>
> A few 1000 should be no problem, we have quite a bit more than that per
> server. As Metin writes, at some point it becomes a really good idea to
> use slashes to namespace them, because you'll run into underlyin
> filesystem issues if you don't. We're taking the first 6 characters of
> the md5 hash of the database name, split those into two 3 character
> strings, and use those as database name prefixes. This may be overkill
> for just a few thousand, but you could employ a more minimal version of
> that strategy to keep the directories manageable. Of course the database
> urls don't get any prettier when you do this.
>
> Note: when using slashes in db names, these should always be encoded to
> %2Fs so the database name your/name/here is actually your%2Fname%2Fhere
> and you'd do well to use that form everywhere, or things can break.
> (Unfortunately, in my view, using slashes *sometimes* works, for
> instance in Futon, which promotes the idea that you can actually use
> them as well. </pet_pieve>)
>
> eric

Thanks for the tips, really helpful.
I will probably use that strategy to partition my data per experiment. 
And use an external watcher script like in couchdb-lounge to replicate 
my common design documents in all databases. And store summary data and 
meta data in a separate database so I can do easily queries spanning 
different dbs.

Tom

Mime
View raw message