Bug 1026352

Summary:	C++ vs Java HotRod client performance
Product:	[JBoss] JBoss Data Grid 6	Reporter:	Radim Vansa <rvansa>
Component:	CPP Client	Assignee:	Tristan Tarrant <ttarrant>
Status:	VERIFIED ---	QA Contact:	Alan Field <afield>
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	6.2.0	CC:	mgencur
Target Milestone:	CR2
Target Release:	6.2.0
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Radim Vansa 2013-11-04 13:45:08 UTC

This bug should track the performance issues for C++ HotRod client.

By preliminary testing we've found that in a test where several hundred clients fire requests to the servers the performance of C++ client is 60% lower than Java client.

Setup:
* 4 cluster nodes, 3 driver nodes
* PUT/GET with 1:2 ratio, 1024 byte entry size
* *NO* delay between requests
* The throughput for > 200 clients (total amount for all nodes) is roughly constant

Results for 6.2.0.ER3.2:
* C++ client: 48k ops/s
* Java client: 120k ops/s

Comment 3 Tristan Tarrant 2013-11-04 20:56:25 UTC

Although I don't believe this can change much, for performance we should compile with the Release profile (ER3.2 is compiled with the Debug profile):

cmake -DHOTROD_JBOSS_HOME=/path/to/jdg-server -DCMAKE_BUILD_TYPE=Release ../path/to/source

Comment 4 Radim Vansa 2014-03-27 09:12:20 UTC

After fixing the bugs in MurmurHash implementation (HRCPP-104) the performance has increased to > 120k ops/s for C++ implementation. Best results for recent Java implementation are 140k ops/s.
These results are acceptable.
Further performance change can be expected when asynchronous operations are implemented.