E1117 11:16:05.656872 46447 StorageClient.inl:123] Request to [10.97.162.206:44500] failed: N6apache6thrift9transport19TTransportExceptionE: AsyncSocketException: write timed out during connection, type = Timed out
140034575853312Host:174170777,latency:1331,e2eLatency:12999
140034575853312Host:174170798,latency:1118,e2eLatency:62702
140034575853312Host:174170679,latency:1425,e2eLatency:65604
140034575853312Host:174170681,latency:1354,e2eLatency:59307
140034575853312Host:174170834,latency:1635,e2eLatency:66143
140034575853312Host:174170685,latency:1531,e2eLatency:65137
140034575853312Host:174170786,latency:1534,e2eLatency:59990
140034575853312Host:174170662,latency:1705,e2eLatency:60732
140034575853312Host:174173810,latency:1412,e2eLatency:69292
140034575853312Host:174170715,latency:1497,e2eLatency:63439
140034575853312Host:174170246,latency:1511,e2eLatency:64306
140034575853312response use time = 1015379 us
140034575853312response completeness = 91%
140034575853312response max latency = 1705 us
这是150并发时延情况，12台机器中一个request超时了，e2e latency也比服务端时延高很多，一次查1500条，每条3kb，每个Meta和Storage Client IOThreadPool都是 12 ，机器是 88core。
-
我理解 关于graphd和storage的线程模型 这个帖子说的，Storage Client 线程模型应该跟Graph Client 类似，并发线程上限跟IOThreadPool大小有关，是不是我的IOThreadPool数目配错了。
-
多台机器起Client可以提高并发吗？，不知道底层Server能抗住吗