Why not running a MR job? It will run a task per region. Le 15 janv. 2013 06:38, "Bing Jiang" <[email protected]> a écrit :
> In my view, if you want to realize Jubtaus, you can use Endpoint. Which > is different from RegionServerObserver. > A client will launch a task about a table, and it will execute an Endpoint > in each region that belongs to that table. > > > 2013/1/15 蔡斌 <[email protected]> > > > > > > > > > Hi We try to build an online machine learning system, like Jubatus( > > > http://www-conf.slac.stanford.edu/xldb2012/talks/xldb2012_wed_LT09_HMakino.pdf > ), > > based on HBase. Our training data will be stored in HBase, and algorithms > > will scan these data to build model. We need a region server level > > co-processor, to scan data stored in that server. Is there any solution > > ready for such requirement? > > ThanksBinCai > > > > > > > > > > > > > -- > Bing Jiang > Tel：(86)134-2619-1361 > weibo: http://weibo.com/jiangbinglover > BLOG: http://blog.sina.com.cn/jiangbinglover > National Research Center for Intelligent Computing Systems > Institute of Computing technology > Graduate University of Chinese Academy of Science >
