Performance Benchmark: Elasticsearch vs Easysearch

Virtual Machine Specifications

Using Alibaba Cloud instance type: ecs.u1-c1m4.4xlarge, PL2: Single disk IOPS performance limit 100,000 (applicable cloud disk capacity range: 461GiB - 64TiB)

vCPU Memory (GiB) Disk (GB) Bandwidth (Gbit/s) Quantity
16 64 500 5000 24

Easysearch Configuration

7-node cluster, version: 1.9.0

Instance Name Internal IP Software vCPU JVM Disk
search-node-01 172.22.75.144 Easysearch 16 31G 500GB
search-node-02 172.23.15.97 Easysearch 16 31G 500GB
search-node-03 172.25.230.228 Easysearch 16 31G 500GB
search-node-04 172.22.75.142 Easysearch 16 31G 500GB
search-node-05 172.22.75.143 Easysearch 16 31G 500GB
search-node-06 172.24.250.252 Easysearch 16 31G 500GB
search-node-07 172.24.250.254 Easysearch 16 31G 500GB

Elasticsearch Configuration

7-node cluster, version: 7.10.2

Instance Name Internal IP Software vCPU JVM Disk
es-node-01 172.24.250.251 Elasticsearch 16 31G 500GB
es-node-02 172.22.75.145 Elasticsearch 16 31G 500GB
es-node-03 172.17.67.246 Elasticsearch 16 31G 500GB
es-node-04 172.22.75.139 Elasticsearch 16 31G 500GB
es-node-05 172.22.75.140 Elasticsearch 16 31G 500GB
es-node-06 172.24.250.253 Elasticsearch 16 31G 500GB
es-node-07 172.24.250.250 Elasticsearch 16 31G 500GB

Monitoring Cluster Configuration

Single-node Easysearch cluster, version: 1.9.0

Instance Name Internal IP Software vCPU Memory Disk
monitor-node-01 172.25.230.226 Monitoring: Console 16 64G 500GB
monitor-node-02 172.23.15.98 Monitoring: Easysearch 16 64G 500GB

Load Generator Configuration

loadgen version: 1.25.0

4 machines for Easysearch, 4 machines for Elasticsearch

Instance Name Internal IP Software vCPU Memory Disk
loadgen-es-01 172.17.67.245 Loadgen - Easysearch 16 64G 500GB
loadgen-es-02 172.22.75.141 Loadgen - Easysearch 16 64G 500GB
loadgen-es-03 172.25.230.227 Loadgen - Easysearch 16 64G 500GB
loadgen-es-04 172.22.75.138 Loadgen - Easysearch 16 64G 500GB
loadgen-el-01 172.24.250.255 Loadgen - Elasticsearch 16 64G 500GB
loadgen-el-02 172.24.251.0 Loadgen - Elasticsearch 16 64G 500GB
loadgen-el-03 172.24.250.248 Loadgen - Elasticsearch 16 64G 500GB
loadgen-el-04 172.24.250.249 Loadgen - Elasticsearch 16 64G 500GB

Test Index Mapping

PUT web_logs
{
  "mappings": {
    "properties": {
      "http_method": {
        "type": "keyword"
      },
      "data_rate": {
        "type": "integer"
      },
      "service": {
        "type": "keyword"
      },
      "client_ip": {
        "type": "ip"
      },
      "memory_consumption": {
        "type": "integer"
      },
      "upstream_duration": {
        "type": "float"
      },
      "request_uri": {
        "type": "keyword"
      },
      "response_length": {
        "type": "integer"
      },
      "request_duration": {
        "type": "float"
      },
      "request_payload_size": {
        "type": "integer"
      },
      "error_status": {
        "type": "keyword"
      },
      "system_metrics": {
        "properties": {
          "queue_length": {
            "type": "integer"
          },
          "memory_allocated": {
            "type": "integer"
          },
          "thread_quantity": {
            "type": "integer"
          },
          "processor_utilization": {
            "type": "integer"
          },
          "active_sessions": {
            "type": "integer"
          }
        }
      },
      "cpu_load": {
        "type": "integer"
      },
      "user_agent_string": {
        "type": "keyword"
      },
      "session_count": {
        "type": "integer"
      },
      "event_time": {
        "type": "date",
        "format": "yyyy-MM-dd'T'HH:mm:ss.SSS"
      },
      "response_code": {
        "type": "integer"
      }
    }
  },
  "settings": {
    "number_of_shards": 7,
    "number_of_replicas": 0,
    "refresh_interval": "30s"
  }
}

Testing Methodology

Every 4 load generators use the bulk write interface to stress test the same cluster's 7 nodes, with each request writing 10,000 documents.

Specific request configuration:

requests:
  - request: #prepare test documents
      method: POST
      runtime_variables:
#        batch_identifier: uuid
      runtime_body_line_variables:
#        routing_identifier: uuid
#      url: $[[env.ES_ENDPOINT]]/_bulk
      url: $[[ip]]/_bulk
      body_repeat_times: 10000
      basic_auth:
       username: "$[[env.ES_USERNAME]]"
       password: "$[[env.ES_PASSWORD]]"
      body: |
        {"index": {"_index": "web_logs", "_type": "_doc", "_id": "$[[uuid]]"}}
        $[[message]]

Test Data Sample

{"http_method":"DELETE","data_rate":1955,"service":"cart-service","client_ip":"120.204.26.240","memory_consumption":1463,"upstream_duration":"1.418","request_uri":"/health","response_length":421,"request_duration":"0.503","request_payload_size":1737,"error_status":"SYSTEM_ERROR","system_metrics":{"queue_length":769,"memory_allocated":1183,"thread_quantity":65,"processor_utilization":68,"active_sessions":837},"cpu_load":70,"user_agent_string":"Mozilla/5.0 (iPad; CPU OS 14_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.1.1","session_count":54,"event_time":"2024-11-16T14:25:21.423","response_code":500}
{"http_method":"OPTIONS","data_rate":10761,"service":"product-service","client_ip":"223.99.83.60","memory_consumption":567,"upstream_duration":"0.907","request_uri":"/static/js/app.js","response_length":679,"request_duration":"1.287","request_payload_size":1233,"error_status":"NOT_FOUND","system_metrics":{"queue_length":565,"memory_allocated":1440,"thread_quantity":148,"processor_utilization":39,"active_sessions":1591},"cpu_load":87,"user_agent_string":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/14.1.1","session_count":354,"event_time":"2024-11-16T05:37:28.423","response_code":502}
{"http_method":"HEAD","data_rate":10257,"service":"recommendation-service","client_ip":"183.60.242.143","memory_consumption":1244,"upstream_duration":"0.194","request_uri":"/api/v1/recommendations","response_length":427,"request_duration":"1.449","request_payload_size":1536,"error_status":"UNAUTHORIZED","system_metrics":{"queue_length":848,"memory_allocated":866,"thread_quantity":86,"processor_utilization":29,"active_sessions":3846},"cpu_load":71,"user_agent_string":"Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)","session_count":500,"event_time":"2024-11-16T15:14:30.424","response_code":403}

Test Results: 1 Primary Shard, 0 Replicas

Elasticsearch Throughput

Elasticsearch Thread and Queue Performance

Resource Consumption

Easysearch Throughput

Easysearch Thread and Queue Performance

Resource Consumption

Comparison

Software Average Cluster Throughput Average Single Node Throughput Maximum Queue Disk Consumption
Elasticsearch 50,000 ops/sec 50,000 ops/sec 811 10GB
Easysearch 70,000 ops/sec 70,000 ops/sec 427 4GB

Test Results: 1 Primary Shard, 1 Replica

Elasticsearch Throughput

Elasticsearch Thread and Queue Performance

Resource Consumption

Easysearch Throughput

Easysearch Thread and Queue Performance

Resource Consumption

Comparison

Software Average Cluster Throughput Average Single Node Throughput Maximum Queue Disk Consumption (~30 million documents)
Elasticsearch 100,000 ops/sec 50,000 ops/sec 791 22GB
Easysearch 140,000 ops/sec 70,000 ops/sec 421 7GB

Test Results: 7 Primary Shards

Elasticsearch Throughput

Elasticsearch Thread and Queue Performance

Resource Consumption

Network

Single node average reception: 26MB/s, corresponding bandwidth: 1456 Mb/s

50 million documents, total storage: 105 GB, single node: 15 GB

Easysearch Throughput

Easysearch Thread and Queue Performance

Resource Consumption

Comparison

Software Average Cluster Throughput Average Single Node Throughput Maximum Queue Disk Consumption
Elasticsearch 350,000 ops/sec 50,000 ops/sec 2449 105GB
Easysearch 600,000 ops/sec 85,714 ops/sec 1172 36GB

Summary

Based on comparative analysis of test results across different scenarios, the following conclusions can be drawn:

  • Easysearch demonstrates significantly improved indexing performance compared to Elasticsearch Easysearch cluster throughput performance improved by 40% - 70%, and the performance improvement effect becomes more significant as the number of shards increases.

  • Easysearch shows substantially better disk compression efficiency compared to Elasticsearch Easysearch cluster disk compression efficiency improved by 2.5 - 3 times, and the compression effect becomes more apparent as data volume increases.

These test results indicate that Easysearch offers superior performance and storage efficiency advantages in log processing scenarios, particularly suitable for large-scale shard and massive data usage scenarios.

Tags: elasticsearch Easysearch performance-testing benchmark search-engine

Posted on Sat, 15 Aug 2026 16:02:11 +0000 by Archangel915

Hot Tags

©2026 Freaks City