On Fri, Jan 6, 2017 at 6:19 AM, Lukáš Vlček <[email protected]> wrote:
> Hi Aaron,
>
> question regarding the Blur data ingestion focus:
>
> Do I read it correctly that Blur is not near-real time system (like both ES
> and Solr are)? For example would Blur be a valid candidate for aggregated
> logging use case? How long does it usually take for indexed data to become
> searchable (ms, sec, mins)?

Blur does suport NRT (like ES & Solr) in a couple ways - take a look
at the various "mutate" methods of the service[1].  Using mutate
you'll see changes directly, for your logging use case, perhaps the
bulkMutate[Start | Add | Finish] would be more appropriate though.

> As for retention of the data what are the strategies to drop old data from
> the index? For example is there anything like dropping old data based on
> index name patters (given the index name contains timestamp)?

Unfortunately right now, it'd be an application-specific approach
[again, using mutations] - there's nothing built-in to age-off old
data.

Thanks,
--tim


[1] - http://incubator.apache.org/blur/docs/0.2.4/Blur.html#Fn_Blur_query

Reply via email to