[
https://issues.apache.org/jira/browse/OAK-9762?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17940338#comment-17940338
]
Alejandro Moratinos commented on OAK-9762:
------------------------------------------
Hi [~angela]
I've done some changes and ran a small set of benchmark comparing to trunk. I
used PrinicipalBasedReadTest (it's not a typo the test is called like that).
All the benchmark runs were executed using the same command:
{code:java}
java -Dwarmup=5 -Druntime=120 -jar target/oak-benchmarks-1.77-SNAPSHOT.jar
benchmark --concurrency 1,3,5,10,20 Oak-Segment-Tar
PrinicipalBasedReadTest{code}
*Branch test runs:*
{noformat}
Run #1
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2911
All paths: 123545
Oak-Segment-Tar 1 26 28 29 31 65
4125 29
Oak-Segment-Tar 3 34 42 48 54 84
7518 48
Oak-Segment-Tar 5 58 71 76 82 139
7850 76
Oak-Segment-Tar 10 81 167 183 197 303
6582 182
Oak-Segment-Tar 20 106 295 361 424 698
6672 360
Run #2
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2953
All paths: 123545
Oak-Segment-Tar 1 27 29 30 32 78
3938 30
Oak-Segment-Tar 3 33 41 47 54 96
7560 48
Oak-Segment-Tar 5 55 70 76 81 122
7936 76
Oak-Segment-Tar 10 98 176 188 200 267
6393 188
Oak-Segment-Tar 20 117 326 381 437 667
6300 380
Run #3
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2848
All paths: 123545
Oak-Segment-Tar 1 26 26 27 29 63
4312 28
Oak-Segment-Tar 3 32 40 45 53 136
7820 46
Oak-Segment-Tar 5 58 72 77 82 130
7794 77
Oak-Segment-Tar 10 105 165 180 196 360
6651 181
Oak-Segment-Tar 20 128 305 362 414 625
6652 361{noformat}
*Trunk test runs:*
{noformat}
Run #1
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2794
All paths: 123545
Oak-Segment-Tar 1 28 28 30 32 97
3994 30
Oak-Segment-Tar 3 35 44 50 56 112
7223 50
Oak-Segment-Tar 5 60 72 78 83 159
7683 78
Oak-Segment-Tar 10 63 159 178 195 337
6749 178
Oak-Segment-Tar 20 103 299 359 415 613
6703 358
Run #2
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2905
All paths: 123545
Oak-Segment-Tar 1 27 29 30 32 76
3985 30
Oak-Segment-Tar 3 34 43 48 54 96
7457 48
Oak-Segment-Tar 5 56 72 77 83 153
7744 77
Oak-Segment-Tar 10 83 155 175 197 409
6807 176
Oak-Segment-Tar 20 126 302 351 399 608
6852 351
Run #3
# PrinicipalBasedReadTest C min 10% 50% 90% max
N mean
Import deep tree: 2917
All paths: 123545
Oak-Segment-Tar 1 28 29 31 33 103
3815 31
Oak-Segment-Tar 3 36 44 50 56 153
7189 50
Oak-Segment-Tar 5 41 69 76 82 169
7912 76
Oak-Segment-Tar 10 100 165 180 195 307
6667 180
Oak-Segment-Tar 20 156 306 364 421 598
6608 364{noformat}
Comparing results I do not see any huge improvement that would worth the risk
to implement these changes on the Principal permission evaluation. Please do
check the changes I've shared in this draft PR
[https://github.com/apache/jackrabbit-oak/pull/2207]
> Implement
> org.apache.jackrabbit.oak.spi.security.authorization.principalbased.impl.AbstractTreePermission.canReadAll
> --------------------------------------------------------------------------------------------------------------------
>
> Key: OAK-9762
> URL: https://issues.apache.org/jira/browse/OAK-9762
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: authorization-principalbased
> Reporter: Angela Schreiber
> Assignee: Angela Schreiber
> Priority: Major
>
> given that _oak-authorization-principalbased_ does not allow for deny
> entries, implementing {{TreePermission.canReadAll}} should be straight
> forward to implement.
> while this won't help with repeated item access with an absolute path, i
> would expect some performance improvements when it comes to traversal (see
> also OAK-9761)
> cc: [~joerghoh], [[email protected]]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
