Datastore JSON API non_transactional commit test

datastore json api test

連續寫入30次，運行在google compute engine上，每次寫入時間大約0.3秒。


$ for i in `seq 30`; do curl -sH "Content-Type: application/json"  -H "Authorization: Bearer xxxxxxxx" https://www.googleapis.com/datastore/v1beta2/datasets/project-id/commit -d @addbody -o /dev/null -w "total:%{time_total}\n"; done
total:0.364
total:0.353
total:0.362
total:0.352
total:0.354
total:0.357
total:0.357
total:0.358
total:0.369
total:0.359
total:0.357
total:0.352
total:0.354
total:0.358
total:0.349
total:0.360
total:0.383
total:0.362
total:0.375
total:0.350
total:0.399
total:0.352
total:0.448
total:0.363
total:0.360
total:0.352
total:0.361
total:0.344
total:0.349
total:0.353



content of @addbody


{
 "mode": "NON_TRANSACTIONAL",
 "mutation": {
  "insertAutoId": [
   {
    "key": {
     "path": [
      {
       "kind": "STAFF"
      }
     ]
    },
    "properties": {
     "name": {
      "stringValue": "Emilie"
     }
    }
   }
  ]
 }
}


runQuery大約0.2~0.3秒

$ for i in `seq 30`; do curl -sH "Content-Type: application/json"  -H "Authorization: Bearer xxxxx" https://www.googleapis.com/datastore/v1beta2/datasets/project-id/runQuery -d @queryBody -o /dev/null -w "total:%{time_total}\n"; done
total:0.287
total:0.276
total:0.268
total:0.291
total:0.285
total:0.272
total:0.291
total:0.276
total:0.282
total:0.328
total:0.275
total:0.280
total:0.281
total:0.270
total:0.270
total:0.276
total:0.281
total:0.343
total:0.274
total:0.313
total:0.278
total:0.282
total:0.275
total:0.285
total:0.284
total:0.358
total:0.275
total:0.283
total:0.287
total:0.273


content of @queryBody

{
 "query": {
  "kinds": [
   {
    "name": "STAFF"
   }
  ],
  "filter": {
   "propertyFilter": {
    "value": {
     "stringValue": "Emilie"
    },
    "property": {
     "name": "name"
    },
    "operator": "equal"
   }
  },
  "limit": 1
 }
}


結論：

1）對create/insert來說速度不快，但query速度挺不錯，如果像Datastore號稱幾百萬幾千萬筆裡面查出100筆，跟幾千筆裡面查出100筆的速度是一樣的話，這0.2~0.3的查詢速度的確很不錯。
2）create/insert可能要做一些operations aggregation的封裝，節省每次commit的時間成本。