uuid - Design approach cassandra for Rowkey and already exists check -

i new cassandra , cassandra giving lot of read timeout errors..tweaked timout still problem may problem design (for application cassandra expected store trillions of data):

question 1 : in cassandra tables use uuid rowkey...but few tables maintainence break rule in user table make email id rowkey....so looking @ tables can understand data stored...is using uuid right approach huge case , second approach user table right or not ???????????

question 2 : have 1 relations table startnodeid, relationtypeid, endnodeid...rowkey uuid relationid.....i define secondary indexes on startnode, relationtype, endnode can have lookup of them business case.........becuase of each new row have check existing relation or not....one approach avoid existing check : take startnodeid, relationtypeid, endnodeid sort them , create hash code , use rowkey...so checking explicitly avoided here..........is right approach ???????

please guide me stuck @ these thoughts...any guidance me

answering first question, until , unless comfortable in handling rowkey non-uuid value, great easier track else go uuid.

regarding second question, why don't try compound key. don't have maintain hashcode stuffs, leave on cassandra.