Ask you a question, very headache:
A big table, assume that t table; There are 36 billion records, probably takes 3.6 T hard disk space;
The t is very simple to insert the data table, the most common way of inserting data;
With the way the t read data table only a very simple way, similar to the select... The from t where a=.. And b=.. Read the concurrency value is small;
That is: the use of t table is very simple; Just is a large amount of data; Take the hard disk is very large;

Now want to use hbase storage, read t table data; Because of the lack of hbase hadoop, zokeeper any project experience, because now need to estimate the hardware conditions, want to consult everybody:
1, need a few computer? What is the function of each computer (such as a zookeeper to use several computers, hadoop use several computers, hbase use several computers, etc.)? Each computer CPU/memory/hard disk?
2, if buy cloud hosting, which company is cheaper, stable?

CodePudding user response:

Consult everybody; Hope to get your guidance;

CodePudding user response:

Suggest directly buy cloud HBase database (ali cloud: https://cn.aliyun.com/product/hbase other vendors parity) don't need to care about the zk, hadoop, HBase cluster details, also ratio you directly buy cheaper ECS build again worry
Recommended two main five from the main (2, 4 g) from (2 t + 4 nuclear 8 g hard disk) (ali cloud about 70000 + year)
From actual usable capacity for node number * single capacity is 5 t/2