thanks Flavio for notificationof yarn,
hope implement in 0.7v
As to nosql or other datbase system, whether set up special extra apdaters tunnel from end
of PE, connected to the PE to recevie PE result data and high performance connect to the distribution
databases or connection pool in S4 based on some configurations(config what database or switch
other databases flexiblilty),
This extra adapters can be defined by different key finder to exists one or some nodes according
to current PE distribution.
post PE definetion, means, when PEs are initialized, it can receive extra different streams,
and not only one onevent function to trig, it can have onevent1, onevent2...
or switch keyfinder during receiving different streams. Just some ideas.
________________________________________
发件人: Flavio Junqueira [fpjunqueira@yahoo.com]
发送时间: 2013年9月6日 19:11
收件人: s4-dev@incubator.apache.org
主题: RE: yarn and nosql support
About yarn, there is work done already, check this out:
https://issues.apache.org/jira/browse/S4-25
About the nosql bit, I'm not entirely sure what you mean with nosql support.
Do you want to write to HBase from a PE for example?
-Flavio
-----Original Message-----
From: Sky Zhao [mailto:sky.zhao@ericsson.com]
Sent: 06 September 2013 11:42
To: s4-dev@incubator.apache.org
Subject: yarn and nosql support
Hi whether should add yarn support and nosql or other storage system
support, has whole solution and integrate old hadoop system.
|