Analyze and optimize the difference between direct OpenSearch queries vs registry API performances

[tloubrieu-jpl]

💡 Description

For example opensearch query, in developer console on AWS:

GET /atm-registry,img-registry,geo-registry,psa-registry,ppi-registry,sbnumd-registry,sbnpsi-registry,rms-registry,naif-registry,en-registry/_search
{
  "query": {
   "bool": {
      "must": [
        {
          "match": {
            "product_class": "Product_Bundle"
          }
        },
        {
          "match": {
            "ops:Tracking_Meta/ops:archive_status": "archived"
          }
        }
      ]
    }
  },
  "fields": [
    "lidvid",
    "ops:Label_File_Info/ops:file_ref",
    "ops:Tracking_Meta/ops:archive_status"
  ],
  "_source": false
}

Result:

{
  "took": 48,
  "timed_out": false,
  "_shards": {
    "total": 0,
    "successful": 0,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": {
      "value": 1935,
      "relation": "eq"
    },
    "max_score": 15.3165,
    "hits": [
      {
        "_index": "sbnumd-registry",
        "_id": "urn:nasa:pds:gbo-ctio::1.0",
        "_score": 15.3165,
        "fields": {
          "lidvid": [
            "urn:nasa:pds:gbo-ctio::1.0"
          ],
...

And API query:

% curl -kv -w '\n* Response time: %{time_total}s\n' 'https://pds.mcp.nasa.gov/api/search/1/classes/bundle?limit=1001&fields=lidvid,ops:Label_File_Info/ops:file_ref,ops:Tracking_Meta/ops:archive_status'
...
* Response time: 1.901252s

⚔️ Parent Epic / Related Tickets

No response

[tloubrieu-jpl]

In this test on chrome we see the request processing takes 1.10s, that is still a lot more that what opensearch only does: