not to be a poor man,hadoop i'am comming

This moment will nap,you will have a dream,But this moment study,you will  interpret a dream.

I leave uncultivatd today,was precisely yesterday perishes tommrrow which person if the body implored.

Thought is already is late,exactly is the earlist time.

Not matter of the today will drag tomrrow.

Time the study pain is temorary,has not  learned the pain islife-long.

Studies this matter,lacks the time,but is lacks diligently.

hadoop bootcamp ,I am coming.

hadoop struture is so many,but we don't need to study all these.we just need learn a fem.

nowdays ,hadoop engineer is fewer.but we need to do more practices .from the begging wo dong't need to set a too high goal.

hadoop is a distributed file system,what's a distributed file system,put your data to different store.what's a hdfs?hdfs is a engine of mapreduce which is composed of jobtracker and tasktrackers.hdfs store all the node data.

to gurantte the data's integrity .namenode's main conception is management and datanode's main conception is data storing.

namenode is commonly called master,and datanode is called slave whose main task is doing jobs.

namemode is usually only one.when it's down,alll the hdfs will be down.but there is copy of namenode.when the master is down,it's copy will replace the master.the size of every data block is 64M,hadoopfs -ls is the hadoop's internal usage.

what's a MR? the output of map is the inputs of reduce,

when we need to dev MR interface,we just need to implements the MR interface.

ssh protocal is the cluster's member's communication basics.

what's the advantage of one time written and more times reading.

ssh=telnet+ftp.

this is a generally introduction of this lesson.

google 'data and accuracy problem.when the data is stored in the ram.search arithmetic is the main task of google or baidu company.