oltpbench_consumer.py

#########################################################################
# From oltpbench/reporting/parsers/summary_parser.py

from oltpbench.reporting.utils import get_value_by_pattern
from oltpbench.reporting.constants import LATENCY_ATTRIBUTE_MAPPING

# NOTE: What we wanted to do was to reuse this function to parse the latency data 
# from the res file as well. 
def parse_latency_data(latency_dict):
    """
    Parses the latency data from the format it is stored in the summary
    file to the format we need for the API request.
    Args:
        latency_dict (dict): The "Latency Distribution" json object in the OLTPBench
        summary file.
    Returns:
        latency (dict): The latency dict required to send to the performance storage
        service.
    """
    latency = {}
    for key, pattern in LATENCY_ATTRIBUTE_MAPPING:
        value = get_value_by_pattern(latency_dict, pattern, None)
        latency[key] = float("{:.4}".format(value)) if value else value
    return latency

#########################################################################
# From oltpbench/reporting/utils.py

def get_value_by_pattern(dict_obj, pattern, default):
    """
    This is similar to .get() for a dict but it matches the
    key based on a substring. This function is case insensitive.
    """
    for key, value in dict_obj.items():
        if pattern.lower() in key.lower():
            return value
    return default

#########################################################################
# From oltpbench/reporting/constants.py

# NOTE: The below array of key value pairs is the problem. It works fine 
# for the summary file but the res file stores the values as lat_min, 
# lat_max, lat_avg. Our solution was just to relax the search string to be
# min, max, av. If you wanted to make it more obvious you could say that
# our original search strings were 25TH_PERCENTILE or something like that. 

# key = API json element name, value = oltpbench output file search string
LATENCY_ATTRIBUTE_MAPPING = [
    ('l_25','25'),('l_75','75'),('l_90','90'), ('l_95','95'), ('l_99','99'),
    ('avg','average'),('median','median'),('min','minimum'), ('max','maximum')]