I plan to deploy 3 Hadoop nodes (2 quadcore CPU, 16GB RAM, 500GB HDD) loaded with data into HDFS. Each data line could be something like:
timestamp;count_of_current_sessions
My aim is to calculate the average count by hour by aggregating, so that the mapreduce result would be:
00:00 -> 50 (between 00:00 and 00:59, there was an average of 50 sessions)
01:00 -> 190
.
.
.
23:00 -> 12
I wish I could use a Hadoop Hive connector with Tableau or JasperSoft to graph the result.
My question is: I've heard about Hadoop latency, and I'm interested in knowing more about the value. For such a set, how long could it be, roughly? Can I use HBase to shorten things?
