##Data stucture in mongodb:
Each document is of this type:
{ _id: encodes time of metric + some random bits, ns: "AWS/EC2"/"AWS/ELB" etc, metric: "CPUUtilization" ts: timestamp dimName: "InstanceId" dimValue: "xyz" }
AWS outputs "Unit" as well, but we will have some conventions and not store it explicitly.
We don't store statistics (i.e. whether it is sum or max or min etc)
##MySQL table for storing fetch configuration client_id: int(11) namespace: varchar(16) metric: varchar(32) dim_name: varchar(32) dim_value: varchar(32) period: int stats: varchar(16) deleted: int created_at: datetime
##Sample feed in api_feed table: { "data": [ {"name": "Sep 1", "cpu": 88}, {"name": "Sep 2", "cpu": 82}, {"name": "Sep 3", "cpu": 84}, {"name": "Sep 4", "cpu": 90}, {"name": "Sep 5", "cpu": 91}, {"name": "Sep 6", "cpu": 86}, {"name": "Sep 7", "cpu": 48} ], "message": "CPU util of abcd machine has been significantly high on Sep7 <a href="http://www.google.com\" target="_blank">See more." }