如何使用过滤器查询优化弹性搜索aggs

Question

When I tried to run this query elasticsearch does not answers, and a lot of cache eviction (related with field cache) occurs.

I do not want to cache any fields in this query, because this is an analyze query which I will run only once a day. Is there any way to get this aggs without using field caching. I tried doc values but it takes 45 second or more. Look forward to your suggestions please.

HEAP Memory : 30GB
ES version 1.3.1
Size of index : 350 GB

{  
   "query": {
      "filtered": {
         "filter": {
            "bool": {
               "must": [
                  {
                     "term": {
                        "type": "161"
                     }
                  },
                  {
                     "term": {
                        "cat": "Math"
                     }
                  }
               ]
            }
         }
      }
   },
   "aggs": {
      "tepo": {
         "terms": {
            "field": "tepo"

         }
      }
   }
}            

Answer 1

You will need to use doc values or field data cache , without which aggregation cant happen. As this query is not very frequent , it is advisable to disable filter cache for that query.

   "filter" : {
        "term" : {
            "cath" : "Math",
            "_cache" : false
        }
    }

This way filter bitset cache wont be generated and you can save some memory on it. LINK - http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/query-dsl-term-filter.html#_caching_18

Another approach would be to limit the amount of field data cache in main memory , and this way you can make sure field data cache wont get over board with the main memory LINK - http://www.elasticsearch.org/guide/en/elasticsearch/guide/current/_limiting_memory_usage.html