I have been doing some research on the HBase and Google's BigTable.
HBase and BigTable look like a massive Matrix store for me.
Each row key is mapped to a big m*n matrix(X), which m is the number of total rows, n is the number of total columns. And X consists tota开发者_如何学Pythonl of Q k*2 matrices. Which m=Q*k, n=2*Q
Is this right?
Thanks!
There are different logical ways of looking at HBase. You certainly can use it as a large sparse matrix. Your description seemed slightly off to me however.
A table is a matrix; rows are composed of column families; column families are composed of column qualifiers. Each column qualifier can have N versions. Practically, you don't want to have more than a small number of column families in the current implementation. So, from a matrix point of view, it is best to think of each table/row as having a single column family.
Also, your definition of X, Q, and k are not clear. Mathematical notation is supposed to be precise, but if you use it without defining your terms, it just makes it impenetrable.
精彩评论