Hello,
I'm currently writing a multi-threaded HBase client using the Java API for test purposes, basically for generating load and data volume on our test cluster. The client can be configured on how many concurrent clients (one thread per client), number of records per client etc ... The thing is, even with only one client (thread), the java process consumes a lot more threads, which is getting beyond 1000 threads when simulating a load with 100 clients with this test application. Data is inserted via batch operations (table.batch(...)). Any ideas on how the HBase Java API uses threads? I only start an explicit thread per simulated client connection in my multi-threaded application. Thanks! Regards, Thomas
