All Products
Search

This Product

    Elasticsearch:Comparison of performance test results between an Elasticsearch V6.8 cluster and an Elasticsearch V8.9 cluster

    Document Center

    Elasticsearch:Comparison of performance test results between an Elasticsearch V6.8 cluster and an Elasticsearch V8.9 cluster

    Last Updated:Mar 26, 2026

    Alibaba Cloud Elasticsearch V8.9 delivers significant improvements over V6.8 in indexing throughput, range query performance, and resource efficiency. This topic compares performance benchmark results between an Alibaba Cloud Elasticsearch V8.9.1 cluster (Contender) and a V6.8.6 cluster (Baseline), using the Rally benchmarking tool from open source Elasticsearch.

    Configurations

    Item Description
    Cluster configurations

    • Version: 8.9.1 and 6.8.6

    • CPU type: Intel

    • Data node specifications: 4 vCPUs and 16 GiB of memory

    • Storage type of data nodes: Enhanced SSD (ESSD)

    • Storage performance level of data nodes: PL1

    • Storage space per data node: 200 GiB

    • Number of data nodes: 3
    Rally configurations Default Rally tracks
    Dataset Built-in dataset http_logs of Rally, with 258,610,096 documents. Get the dataset from Elasticsearch Rally Hub.
    Number of shards

    • Primary shards: 6

    • Replica shards: 0
    bulk_size 2,000 documents submitted each time
    bulk_indexing_clients 10 clients used in parallel

    Test results

    These results are for reference only. Run stress tests with your own business production data before making upgrade decisions.

    How to read these results

    Before reading the metric tables, note the following:

    • Check error rate first. If the error rate is non-zero, the other metrics are not meaningful. All tasks in this test report 0% error rate.

    • Evaluate in order: error rate → throughput → latency percentiles.

    • Rally-specific latency definitions differ from tools such as JMeter:

      • Latency: time from when a request is submitted to when the complete response is received, including any queue waiting time before Elasticsearch begins processing.

      • Service time: time from when Elasticsearch starts processing a request to when the response is received, excluding queue waiting time. A large gap between latency and service time indicates requests are spending significant time waiting in queue.

    • Cumulative timing metrics (indexing time, merge time, refresh time, flush time) are cumulative totals across the test run, not wall-clock time.

    • Baseline = V6.8.6 cluster. Contender = V8.9.1 cluster. A negative Diff % on latency metrics means improvement (lower is faster). A positive Diff % on throughput means improvement (higher is faster).

    Main metrics

    Elasticsearch cluster Indexing rate Indexing latency
    V8.9.1 cluster <img> <img>
    V6.8.6 cluster <img> <img>

    Test report

    For metric definitions, see Metrics and task parameters for performance stress testing.

    Metric Task Baseline Contender Diff Unit Diff %
    Cumulative indexing time of primary shards - 195.859 129.703 -66.1556 min -33.78%
    Min cumulative indexing time across primary shard - 8.33E-05 0 -8.00E-05 min -100.00%
    Median cumulative indexing time across primary shard - 1.42498 0.857517 -0.56747 min -39.82%
    Max cumulative indexing time across primary shard - 26.5076 16.5645 -9.94305 min -37.51%
    Cumulative indexing throttle time of primary shards - 0 0 0 min 0.00%
    Min cumulative indexing throttle time across primary shard - 0 0 0 min 0.00%
    Median cumulative indexing throttle time across primary shard - 0 0 0 min 0.00%
    Max cumulative indexing throttle time across primary shard - 0 0 0 min 0.00%
    Cumulative merge time of primary shards - 119.628 50.4829 -69.1454 min -57.80%
    Cumulative merge count of primary shards - 1562 601 -961 - -61.52%
    Min cumulative merge time across primary shard - 0 0 0 min 0.00%
    Median cumulative merge time across primary shard - 0.146967 0.081167 -0.0658 min -44.77%
    Max cumulative merge time across primary shard - 19.4382 9.88038 -9.55778 min -49.17%
    Cumulative merge throttle time of primary shards - 80.369 15.6503 -64.7187 min -80.53%
    Min cumulative merge throttle time across primary shard - 0 0 0 min 0.00%
    Median cumulative merge throttle time across primary shard - 0.036 0 -0.036 min -100.00%
    Max cumulative merge throttle time across primary shard - 13.8056 4.30382 -9.50173 min -68.83%
    Cumulative refresh time of primary shards - 12.5059 17.8214 5.31553 min 42.50%
    Cumulative refresh count of primary shards - 13293 6140 -7153 - -53.81%
    Min cumulative refresh time across primary shard - 0.00025 0 -0.00025 min -100.00%
    Median cumulative refresh time across primary shard - 0.086067 0.119483 0.03342 min 38.83%
    Max cumulative refresh time across primary shard - 1.50832 2.15253 0.64422 min 42.71%
    Cumulative flush time of primary shards - 1.24528 3.80468 2.5594 min 205.53%
    Cumulative flush count of primary shards - 129 817 688 - 533.33%
    Min cumulative flush time across primary shard - 0 3.33E-05 3.00E-05 min 0.00%
    Median cumulative flush time across primary shard - 0.000567 0.016117 0.01555 min 2744.12%
    Max cumulative flush time across primary shard - 0.220317 0.5159 0.29558 min 134.16%
    Total Young Gen GC time - 71.731 17.58 -54.151 s -75.49%
    Total Young Gen GC count - 2609 922 -1687 - -64.66%
    Total Old Gen GC time - 0 0 0 s 0.00%
    Total Old Gen GC count - 0 0 0 - 0.00%
    Store size - 20.5377 18.1059 -2.43177 GB -11.84%
    Translog size - 15.5092 0.000209 -15.509 GB -100.00%
    Heap used for segments - 96.9595 0 -96.9595 MB -100.00%
    Heap used for doc values - 0.143082 0 -0.14308 MB -100.00%
    Heap used for terms - 83.811 0 -83.811 MB -100.00%
    Heap used for norms - 0.041382 0 -0.04138 MB -100.00%
    Heap used for points - 5.80011 0 -5.80011 MB -100.00%
    Heap used for stored fields - 7.16398 0 -7.16398 MB -100.00%
    Segment count - 688 56 -632 - -91.86%
    Total Ingest Pipeline count - 0 0 0 - 0.00%
    Total Ingest Pipeline time - 0 0 0 ms 0.00%
    Total Ingest Pipeline failed - 0 0 0 - 0.00%
    Min Throughput index-append 144562 207170 62608.3 docs/s 43.31%
    Mean Throughput index-append 150015 219531 69516.8 docs/s 46.34%
    Median Throughput index-append 148913 217484 68571.1 docs/s 46.05%
    Max Throughput index-append 155476 238529 83052.4 docs/s 53.42%
    50th percentile latency index-append 118.389 80.1159 -38.2727 ms -32.33%
    90th percentile latency index-append 181.669 123.321 -58.3479 ms -32.12%
    99th percentile latency index-append 595.738 394.778 -200.96 ms -33.73%
    99.9th percentile latency index-append 924.456 2125.09 1200.63 ms 129.87%
    99.99th percentile latency index-append 1329.83 2816.56 1486.73 ms 111.80%
    100th percentile latency index-append 2490.11 3341.14 851.03 ms 34.18%
    50th percentile service time index-append 118.389 80.1159 -38.2727 ms -32.33%
    90th percentile service time index-append 181.669 123.321 -58.3479 ms -32.12%
    99th percentile service time index-append 595.738 394.778 -200.96 ms -33.73%
    99.9th percentile service time index-append 924.456 2125.09 1200.63 ms 129.87%
    99.99th percentile service time index-append 1329.83 2816.56 1486.73 ms 111.80%
    100th percentile service time index-append 2490.11 3341.14 851.03 ms 34.18%
    error rate index-append 0 0 0 % 0.00%
    Min Throughput default 7.95637 19.9543 11.9979 ops/s 150.80%
    Mean Throughput default 7.96001 19.9645 12.0045 ops/s 150.81%
    Median Throughput default 7.96018 19.9661 12.0059 ops/s 150.82%
    Max Throughput default 7.96335 19.9692 12.0058 ops/s 150.76%
    50th percentile latency default 90.2508 6.14099 -84.1098 ms -93.20%
    90th percentile latency default 104.465 7.43625 -97.0288 ms -92.88%
    99th percentile latency default 120.173 35.009 -85.1635 ms -70.87%
    100th percentile latency default 122.049 73.9764 -48.0721 ms -39.39%
    50th percentile service time default 88.8955 4.93161 -83.9639 ms -94.45%
    90th percentile service time default 103.013 6.18375 -96.8297 ms -94.00%
    99th percentile service time default 117.761 25.6137 -92.1469 ms -78.25%
    100th percentile service time default 121.368 73.0867 -48.2812 ms -39.78%
    error rate default 0 0 0 % 0.00%
    Min Throughput term 49.547 42.1405 -7.40655 ops/s -14.95%
    Mean Throughput term 49.5671 42.4414 -7.12567 ops/s -14.38%
    Median Throughput term 49.5671 42.4414 -7.12567 ops/s -14.38%
    Max Throughput term 49.5872 42.7424 -6.84478 ops/s -13.80%
    50th percentile latency term 10.1058 8.54857 -1.55724 ms -15.41%
    90th percentile latency term 11.9156 10.4358 -1.47983 ms -12.42%
    99th percentile latency term 34.4109 12.6224 -21.7885 ms -63.32%
    100th percentile latency term 43.6123 22.1556 -21.4567 ms -49.20%
    50th percentile service time term 9.07766 7.13231 -1.94535 ms -21.43%
    90th percentile service time term 10.1664 8.7835 -1.38285 ms -13.60%
    99th percentile service time term 19.2223 11.7613 -7.46097 ms -38.81%
    100th percentile service time term 42.2021 20.9707 -21.2314 ms -50.31%
    error rate term 0 0 0 % 0.00%
    Min Throughput range 1.00138 24.8781 23.8768 ops/s 2384.39%
    Mean Throughput range 1.00191 24.9019 23.9 ops/s 2385.45%
    Median Throughput range 1.00184 24.9036 23.9018 ops/s 2385.79%
    Max Throughput range 1.00273 24.9222 23.9194 ops/s 2385.42%
    50th percentile latency range 461.013 8.53625 -452.477 ms -98.15%
    90th percentile latency range 484.002 10.119 -473.883 ms -97.91%
    99th percentile latency range 570.831 87.6618 -483.169 ms -84.64%
    100th percentile latency range 575.619 111.375 -464.244 ms -80.65%
    50th percentile service time range 459.381 7.46765 -451.914 ms -98.37%
    90th percentile service time range 482.296 9.07516 -473.221 ms -98.12%
    99th percentile service time range 568.99 29.1409 -539.849 ms -94.88%
    100th percentile service time range 574.188 110.444 -463.744 ms -80.77%
    error rate range 0 0 0 % 0.00%
    Min Throughput hourly_agg 0.200542 0.200569 3.00E-05 ops/s 0.01%
    Mean Throughput hourly_agg 0.20075 0.200786 4.00E-05 ops/s 0.02%
    Median Throughput hourly_agg 0.200721 0.200756 3.00E-05 ops/s 0.02%
    Max Throughput hourly_agg 0.201076 0.201128 5.00E-05 ops/s 0.03%
    50th percentile latency hourly_agg 1763.18 1924.92 161.747 ms 9.17%
    90th percentile latency hourly_agg 1851.01 2009.28 158.272 ms 8.55%
    99th percentile latency hourly_agg 2020.41 2118.9 98.4938 ms 4.87%
    100th percentile latency hourly_agg 2021.06 2302.15 281.092 ms 13.91%
    50th percentile service time hourly_agg 1760.45 1921.89 161.444 ms 9.17%
    90th percentile service time hourly_agg 1848.91 2007.78 158.876 ms 8.59%
    99th percentile service time hourly_agg 2019.07 2116.37 97.3059 ms 4.82%
    100th percentile service time hourly_agg 2019.19 2300.89 281.704 ms 13.95%
    error rate hourly_agg 0 0 0 % 0.00%
    Min Throughput scroll 24.9909 25.0461 0.05527 pages/s 0.22%
    Mean Throughput scroll 24.995 25.076 0.08104 pages/s 0.32%
    Median Throughput scroll 24.9954 25.0691 0.07377 pages/s 0.30%
    Max Throughput scroll 24.997 25.1378 0.14077 pages/s 0.56%
    50th percentile latency scroll 482.33 355.086 -127.245 ms -26.38%
    90th percentile latency scroll 506.714 373.589 -133.125 ms -26.27%
    99th percentile latency scroll 570.072 407.765 -162.306 ms -28.47%
    100th percentile latency scroll 662.62 443.031 -219.589 ms -33.14%
    50th percentile service time scroll 480.324 353.174 -127.15 ms -26.47%
    90th percentile service time scroll 505.168 371.193 -133.975 ms -26.52%
    99th percentile service time scroll 567.947 405.43 -162.517 ms -28.61%
    100th percentile service time scroll 660.957 440.533 -220.424 ms -33.35%
    error rate scroll 0 0 0 % 0.00%