[ https://issues.apache.org/jira/browse/HDDS-3223?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17060612#comment-17060612 ]
runzhiwang edited comment on HDDS-3223 at 3/18/20, 12:54 PM: ------------------------------------------------------------- I think it's related to our own operating system. was (Author: yjxxtd): I think it's related to the User-Agent version. when the version is aws-cli/1.17.13 Python/2.7.5 botocore/1.14.13, read is slow, but when the version is Boto3/1.12.22 Python/3.6.8 Botocore/1.15.22 read is fast. > Read a big object cost 2 times more than write it by s3g > -------------------------------------------------------- > > Key: HDDS-3223 > URL: https://issues.apache.org/jira/browse/HDDS-3223 > Project: Hadoop Distributed Data Store > Issue Type: Improvement > Reporter: runzhiwang > Assignee: runzhiwang > Priority: Major > Attachments: screenshot-3.png, screenshot-5.png > > > By s3gateway, write a 187MB file cost 5 seconds, but read it cost 17 seconds. > Both write and read will split 187MB file into 24 parts, so write/read has 24 > POST/GET requests, I find s3g process the first 10 GET requests in parallel > and process the next 14 GET requests in sequential. I use {code:java}tcpdump > -i eth0 -s 0 -A 'tcp dst port 9878 and tcp[((tcp[12:1] & 0xf0) >> 2):4] = > 0x47455420' -w read.cap{code} to capture the GET request to s3gateway , as > the first image shows. The first 10 GET requests range from 3.54 second to > 3.56 second. But the next 14 GET requests range from 4.41 second to 12.23 > second. I also capture the PUT request to s3gateway, as the second image > shows, the 24 PUT requests range from 0.63 second to 3.48 second, that's the > reason why write is faster than read. I think the reason is in aws-cli. I > will continue to find it out. > !screenshot-3.png! > !screenshot-5.png! -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org