questions about HBase, data export and log replication

[隐山堂 <79...@qq.com>]

Hi there, 


1.  I understand there are 3 types of tables in Trafodion, HBase, trafodion and Hive. Are all col in trafodion table being stored in ONE col. family?
2. Let's say some of my manufacturing enterprise clients is using HIVE/HBase and is looking for more SQL-Friendly reporting and medium-to-low concurrency OLTP applications, the response time and concurrency requirement isn't very high. I notice Trafodion is able to visit data stored in HBase/HIVE, i.e. select * from hbase.”_ROW_”.”table1”; Under what circumstances, it's better to move data from HBase/HIVE to Trafodion tables?  For example, to replace HIVE for offline reports, or replacing HBase for low-concurrency order management scenarios,  do I need to export from my current HBase+Hive?  
If exporting is needed, is there a tool or document with step-by-step details? The reason I'm asking this is because in some bidding processes, we have very limited time to run through a series of standard tests to test FUNCTIONALITIES, e.g. ANSI SQL compatibility etc. Performance won't be tested, but we need to finish these functional tests quickly. So, how fast we could move table from client's sample HBase/HIVE sources into Trafodoin tables is very important.  Any guidance will be appreciated. 
3. Trafodion's architecture shows MXOSRVR for each SQL statement and DCS Master to allocation client connections to MXOSRVR. Let's say I'm running CDH 5.4.8, Centos 6.5, 2.1G CPU and 64G RAM, how do I estimate how much concurrencies I could handle with a 6-node cluster, in IoT scenarios, with 50,000-100,000 events/sec realtime injection from sensor devices, how many adhoc and/or reporting queries can I run over 100M records? 
4. what kind of protocol do you replicate logs across a distributed system, something similar to Raft? Quorum? 


Thanks!


Best, 
Felix