Is Riak the right software for me?

Philip flips01 at googlemail.com
Sun Mar 4 08:54:58 EST 2012


It looks like these high wait times occur when the request hits a node
which does not store the requested file. The node will download the
whole file into memory and after that he will send the data to the
client. How about directly pipe'ing the download to the output stream
or (even better) redirect the client to a node with the file so the
overall network usage decreases?

Am 4. März 2012 11:40 schrieb Philip <flips01 at googlemail.com>:
> I am looking for a software to host millions of files in the order of
> 2-20MB for web serving. A few (~5%) files are "hot" and accessed
> heavily but most of the files are cold. The system is going to work
> like a big cache and therefore I need to make sure that the cluster
> won't go out of space while new files are beeing added. I'd like to
> store the last time a file was accessed and delete the least accessed
> files to make sure there is always 20-30% free space in the cluster.
> This system is currently spread across alot of different standalone
> servers and serves a few gbit of bandwidth.
>
> Is this possible with riak? Is there anything I should take into consideration?
>
> I did a little testing with a 4 node cluster. Basically with the
> default configuration, I've only changed the ip addresses and disabled
> nagle buffering for http. When I want to access a 20MB file through
> the REST API it sometimes takes 4-8 seconds until any data is beeing
> sent to the client. I've made use of the "r" parameter (r=1) and the
> wait time improved a little bit but is still unacceptable for
> production use. Is there some config option I am missing here to get
> the requested file without such high wait times?
>
> Best Regards
> Philip




More information about the riak-users mailing list