diff --git a/TECH. md b/TECH. md
new file mode 100644
index 000000000..9df85715a
--- /dev/null
+++ b/TECH. md	
@@ -0,0 +1,33 @@
+
+## Elastic search
+
+Elasticsearch is used to retrieve data used in many pages. When data is
+created, updated, or deleted, and index of this data is saved into elastic
+search. This is denormlised data, which for us means it's saved into elastic
+search in a structure ready to be displayed onto our pages. Data in postgres,
+on the other hand, is saved in a different structure, that ensures data
+integrity.
+
+For example, a crop needs a default photo. The photo used is the most "liked"
+photo of plantings, seeds or harvests for that crop. To get this from
+postgresql you'd need to look up the crops tables, the
+plantings,seeds,harvests, the photos, and then the likes. Instead of doing
+that on every display of a crop photo, we calculate this at data change and
+saved into the Elasticsearch index for crops.
+
+### Installing Elasticsearch
+
+Currently we use elastic search 7. You can check if this is still true by
+looking for the `ELASTIC_SEARCH_VERSION` variable in the `.travis.yml` file
+
+To install ES on a debian/ubuntu machine you can use the same script used in
+travis-ci:
+
+```
+export ELASTIC_SEARCH_VERSION="7.5.1-amd64" ./script/install_elasticsearch.sh
+```
+
+You can check that it started happily using:
+```
+./script/check_elasticsearch.sh
+```