storm-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Patrick Wiener <patrick.wie...@web.de>
Subject Re: MongoDB + Storm
Date Tue, 30 Dec 2014 10:47:16 GMT
Thank you very much - spatial4j seems to be the right choice. 


> Am 30.12.2014 um 04:21 schrieb Michael Rose <michael@fullcontact.com>:
> 
> Check out spatial4j for in memory fencing.
> 
> On Monday, December 29, 2014, Elliott Bradshaw <ebradshaw1@gmail.com <mailto:ebradshaw1@gmail.com>>
wrote:
> For spatial lookups I would advise against calling out to MongoDB unless your set of
geometries is very large.  It would likely be better to do this processing in memory on the
storm cluster.  The Java Topology Suite would be a good start.
> 
> On Dec 29, 2014 4:10 PM, "Tero Paananen" <teropaananen@gmail.com <javascript:_e(%7B%7D,'cvml','teropaananen@gmail.com');>>
wrote:
> > I don’t actually know how i could cache the data to perform this sort of mongoDB
query ($geoWithin) on cached entries?!
> 
> I was thinking of an API that would allow you to perform geoWithin
> queries on an in-memory collection of geoJSON polygons. Essentially an
> in-memory geospatial database of sorts. I have no idea if one exists,
> but if it does, that would most definitely be the fastest way to do
> those lookups.
> 
> Something else came to my mind after I wrote the last post. I had very
> good experience in using Elasticsearch for this very same purpose, and
> while looking into solutions in this space at the time we were
> implementing our solutions a lot of people were recommending
> Postgresql based solutions. Those might be worth looking into as well.
> 
> -TPP
> 
> 
> -- 
> Michael Rose (@Xorlev <https://twitter.com/xorlev>)
> Senior Platform Engineer, FullContact <http://www.fullcontact.com/>
> michael@fullcontact.com <mailto:michael@fullcontact.com>


Mime
View raw message