[
https://issues.apache.org/jira/browse/IOTDB-4449?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jinrui Zhang reassigned IOTDB-4449:
-----------------------------------
Assignee: Yukun Zhou (was: Jinrui Zhang)
> [ MultiLeaderConsensus ] Very low write performance at 100k timeseries per dev
> ------------------------------------------------------------------------------
>
> Key: IOTDB-4449
> URL: https://issues.apache.org/jira/browse/IOTDB-4449
> Project: Apache IoTDB
> Issue Type: Bug
> Components: mpp-cluster
> Affects Versions: 0.14.0-SNAPSHOT
> Reporter: 刘珍
> Assignee: Yukun Zhou
> Priority: Major
> Attachments: 1_dev_10w_sensor.conf, image-2022-09-19-18-08-27-560.png
>
>
> m_0918_a28329e
> multiLeader，{color:#DE350B}*在单个设备包含的序列较多时，写入性能变得非常慢*{color}。
> 1用户，1dev，10w序列，batch size=2，LOOP=10，即每序列写入20个点，相同数据库配置
> 与新单机的性能对比：
> !image-2022-09-19-18-08-27-560.png!
> 测试环境
> 1. 192.168.10.71/72/73/74 48核384G
> bm在ip71
> confignode配置参数
> MAX_HEAP_SIZE="8G"
> schema_region_consensus_protocol_class=org.apache.iotdb.consensus.ratis.RatisConsensus
> data_region_consensus_protocol_class=org.apache.iotdb.consensus.multileader.MultiLeaderConsensus
> schema_replication_factor=3
> data_replication_factor=3
> connection_timeout_ms=3600000
> DataNode配置
> MAX_HEAP_SIZE="256G"
> MAX_DIRECT_MEMORY_SIZE="32G"
> max_connection_for_internal_service=300
> enable_timed_flush_seq_memtable=true
> seq_memtable_flush_interval_in_ms=3600000
> seq_memtable_flush_check_interval_in_ms=600000
> enable_timed_flush_unseq_memtable=true
> unseq_memtable_flush_interval_in_ms=3600000
> unseq_memtable_flush_check_interval_in_ms=600000
> max_deduplicated_path_num=1000000
> max_waiting_time_when_insert_blocked=3600000
> query_timeout_threshold=36000000
> 2. bm配置见附件
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
