[Photon] performance of edgngram

[Christoph Lingg]

Hi guys,

I am tweaking the performance of the komoot fork of photon and asked the mailing list for some help: https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/elasticsearch/bjl2PJEhYsg/FSb5B52C24YJ

There is an interesting statement I wanted to share with you:
> I had some pretty weird result in the past where under specific circonstances I had better performance results with prefix queries than edgengram with a huge volume of data.

So it will make sense to give it a try. Btw. Felix is exporting a json dump so we can soon start to play with world-wide data.

Cheers,
Christoph