[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-02-16 Thread Alexey Popov (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16366954#comment-16366954
 ] 

Alexey Popov commented on IGNITE-7191:
--

No improvement is observed. Closing the ticket

> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
>Priority: Major
> Fix For: 2.5
>
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-19 Thread Alexey Popov (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16332465#comment-16332465
 ] 

Alexey Popov commented on IGNITE-7191:
--

[~ptupitsyn] , I run some benchmarks and check wireshark logs with different 
socket size options.

0 (default) works as ~ 256k with a good network at Win client and Linux server. 
So, it seems that this improvement is not required for modern OSs.

I will double check the results later.

> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
>Priority: Major
> Fix For: 2.5
>
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-19 Thread Vladimir Ozerov (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16331912#comment-16331912
 ] 

Vladimir Ozerov commented on IGNITE-7191:
-

[~ptupitsyn], [~alexey.tank2], Moving to 2.5 because performance impact should 
be evaluated carefully.

> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
>Priority: Major
> Fix For: 2.5
>
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-11 Thread Pavel Tupitsyn (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16322255#comment-16322255
 ] 

Pavel Tupitsyn commented on IGNITE-7191:


[~apopov] why 64k and not 128k or anything else? Did you perform any benchmarks?

> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
> Fix For: 2.4
>
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-11 Thread Alexey Popov (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16322235#comment-16322235
 ] 

Alexey Popov commented on IGNITE-7191:
--

[~ptupitsyn] and [~tledkov-gridgain], could you please review the changes

> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
> Fix For: 2.4
>
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-11 Thread Alexey Popov (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16322232#comment-16322232
 ] 

Alexey Popov commented on IGNITE-7191:
--

Scope of changes:
1. Server side default socket size for thin clients
2. JDBC thin default socket size & tests
3. .Net thin client socket size & tests

PR:
https://github.com/apache/ignite/pull/3353

TC:
https://ci.ignite.apache.org/project.html?projectId=IgniteTests2023Java7_IgniteTests2023Java7=pull%2F3353%2Fhead


> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)


[jira] [Commented] (IGNITE-7191) JDBC: set socket buffer to 64k by default

2018-01-10 Thread ASF GitHub Bot (JIRA)

[ 
https://issues.apache.org/jira/browse/IGNITE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16320509#comment-16320509
 ] 

ASF GitHub Bot commented on IGNITE-7191:


GitHub user apopovgg opened a pull request:

https://github.com/apache/ignite/pull/3353

IGNITE-7191 JDBC: set socket buffer to 64k by default



You can merge this pull request into a Git repository by running:

$ git pull https://github.com/gridgain/apache-ignite ignite-7191

Alternatively you can review and apply these changes as the patch at:

https://github.com/apache/ignite/pull/3353.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

This closes #3353


commit 5f6bd3e2fa0567160031cdd067b3e2b1f20c9972
Author: apopov 
Date:   2018-01-10T16:08:05Z

IGNITE-7191 JDBC: set socket buffer to 64k by default




> JDBC: set socket buffer to 64k by default
> -
>
> Key: IGNITE-7191
> URL: https://issues.apache.org/jira/browse/IGNITE-7191
> Project: Ignite
>  Issue Type: Improvement
>  Components: jdbc
>Affects Versions: 2.1
>Reporter: Alexey Popov
>Assignee: Alexey Popov
>
> TCP socket buffer size can significantly affect user-visible performance 
> (actually, latency) for SQL queries, especially if Ignite cluster is run at 
> the remote hosts. 
> It is better to have 64k TCP socket buffer size (instead of default 8k) by 
> default to avoid possible delays in TCP transport.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)