Spot bid broker

This is a python script that spawns EC2 spot instances for minimum bid and maximum uptime(specific instance type, region and bid).

Go through the following steps to place the spot bids:

1. Using pip, install boto, mock, nose and wsgiref

2. Add your aws keys, region name, ami, key pair name, security group name to the config file

3. First, retrieve the 2-month spot history by executing the following command: $ python spot_history_data.py <instance_type> > results.txt

4. Run main.py to place the spot bid with max uptime: $ python main.py start

---

Details about functions:

parse_regions(): The spot history data contains data pertaining to all availability zones. This functions splits data based on availability zones.

pad_region(): Used to pad data to make an hourly data(For easier analysis)

simplify_data(): Display data as a tuple - <hour,price>

stats(): To compute Avg uptime, Avg. Downtime, Availability of a persistent request for bids varying from 0 to on-demand cost.

get_recent_history(): Get mean and standard deviation of the data.

auto_corr(): Compute the autocorrelation of the spot pricing history with a lag factor varying from 0hrs to the input provided(mostly 50-60hrs)

get_stl(): Makes decision on which availability zone to choose by comparing standard deviations of residual graphs(by removing seasonality and trend)

read_user_data_from_local_config(): Read keys from the local config file

get_spot_price_history(): Get spot history data from AWS.

create_client(): Create an EC2 client session using boto

provision_instance(): Place a spot request with given parameters.

wait_for_up(): Used to connect to the spot instance and get its public IP

Credits

Some part of the code comes from spot-ec2-proxifier project