如何在heroku上托管的rails项目中实现弹性搜索？

My plan is to run elastic search on a separate linux server as a web service. Then I will make calls to it directly from my app to search with strings or other unstructured data. What I don't understand is will this basically replace my database for certain objects or at the time that the object is created, should I index it's ID, object type and tags in elastic search as well as simultaneously save the entire object to my database? Then I can make a call to the web service to get the object's ID and then query my database based on that ID to retrieve the actual object?My plan is to run elastic search on a separate