Using coordinates, summarizing numeric field in different docs, regional granularities, regional graphs, etc


(Dieter Plaetinck) #1

Hey all,
I've been playing with elasticsearch and kibana for a few smaller projects
and am quite satisfied,
however now I want to do something a little more complicated so I'ld like
to double check it is indeed possible.

imagine you run a website, and users are uploading content all the time,
from anywhere in the world, to one of your few datacenters.
for every finished upload, lets log a document with the following fields:

  • datetime of upload
  • coordinates
  • datacenter_key (string with about 10 possible values)
  • upload_speed (numeric)

Can we do the following?:

  • See on a map how the average upload speed differs geographically, by
    using data of (say) the last 24h or the last week (and averaging the
    upload_speed value over all those documents! not just counting how many
    documents)
    Can we use the coordinates data to automatically divide into regions
    (they don't have to map to political regions) or is it mandatory to use a
    country/state code?
    Also, is it possible to add the regions together and see a more high
    level overview of the map? (say north america, south america, africa,
    europe, etc etc)

  • let's say I want to see how the above evolves over time, so I want to see
    a graph that tells me how the average upload speed (calculated over all
    docs) evolved over time. We should be able to do this:

    • globally (i.e. for all documents, irrespective of coordinates/country)
    • for "high level regions" (north america, south america, africa,
      europe, etc etc)
    • for smaller regions. it seems there's about 250 countries + states in
      the world, so maybe 25 graphs of 10 regions each. but again they don't
      necessarily have to be political regions, if there's a way to come up with,
      say 30 regions that cover the globe,
      that would be fine too.
  • get a graph per datacenter key how the average upload speed evolves over
    time.

  • Everything above, but upper 90th percentile instead of average.

thanks!
Dieter

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


(system) #2