You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@shardingsphere.apache.org by 吴晟 Sheng Wu <wu...@foxmail.com> on 2019/02/13 03:55:59 UTC

Database metric and slow sql have landed in SkyWalking master branch

Hi


We have discussed about the integration of metric for database before.
Yesterday, my pr[1] got passed and landed in SkyWalking master branch. 
By that, SkyWalking could provide 
1. Identify database from trace
2. Analysis trace metrics, including call per minutes, avg response time, p99/95/90/75/50 response time and success rate
3. Sample TopN database statements by latency, including SQL of course.


All these, including the incoming UI, will be released in SkyWalking 6.1


So, for sharding-JDBC, even you plugin provide[2], which definitely haven't, you will not be recognized as a logic database, because all real database access are through JDBC to real database. So if you could have metric and slow statement for logic database, you should provide another plugin to report database register and telemetry report. I really hope you could be another source to SkyWalking oap platform. Right now, we have telemetry sources from multiple language agent(java from SkyWalking, .net, php, nodejs from SkyAPM), Istio, Zipkin. Envoy is on the way.
For Sharding-Proxy, that may be better, since at the client side(user app side), the proxy would be recognized as a logic database. But I am not sure, because have no one talk about this before.


I hope you could consider how to deal with the metric and observability for the project, which would be very important for end user.




[1] https://github.com/apache/incubator-skywalking/pull/2239
[2] https://github.com/apache/incubator-skywalking/pull/1934


------------------
Sheng Wu
Apache SkyWalking, ShardingSphere, Zipkin
Twitter, wusheng1108

Re: Database metric and slow sql have landed in SkyWalking master branch

Posted by 吴晟 Sheng Wu <wu...@foxmail.com>.
Analysis trace metric, I mean analysis database access metrics from trace.
Sorry.



Sheng Wu
Apache SkyWalking, ShardingSphere, Zipkin

From Wu Sheng 's phone.


------------------ Original ------------------
From: 吴晟 Sheng Wu <wu...@foxmail.com>
Date: Wed,Feb 13,2019 11:56 AM
To: dev <de...@shardingsphere.apache.org>
Subject: Re: Database metric and slow sql have landed in SkyWalking master branch



Hi


We have discussed about the integration of metric for database before.
Yesterday, my pr[1] got passed and landed in SkyWalking master branch. 
By that, SkyWalking could provide 
1. Identify database from trace
2. Analysis trace metrics, including call per minutes, avg response time, p99/95/90/75/50 response time and success rate
3. Sample TopN database statements by latency, including SQL of course.


All these, including the incoming UI, will be released in SkyWalking 6.1


So, for sharding-JDBC, even you plugin provide[2], which definitely haven't, you will not be recognized as a logic database, because all real database access are through JDBC to real database. So if you could have metric and slow statement for logic database, you should provide another plugin to report database register and telemetry report. I really hope you could be another source to SkyWalking oap platform. Right now, we have telemetry sources from multiple language agent(java from SkyWalking, .net, php, nodejs from SkyAPM), Istio, Zipkin. Envoy is on the way.
For Sharding-Proxy, that may be better, since at the client side(user app side), the proxy would be recognized as a logic database. But I am not sure, because have no one talk about this before.


I hope you could consider how to deal with the metric and observability for the project, which would be very important for end user.




[1] https://github.com/apache/incubator-skywalking/pull/2239
[2] https://github.com/apache/incubator-skywalking/pull/1934


------------------
Sheng Wu
Apache SkyWalking, ShardingSphere, Zipkin
Twitter, wusheng1108