def generate_graphs(self, show=False):
filename = "task_arrival_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
generate_cdf(self.df, "ts_submit", os.path.join(self.folder, filename),
"Time (ms)", "Num. tasks (CDF)", show)
return filename
def generate_graphs(self, show=False):
filename = "job_runtime_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
generate_cdf(self.df, "critical_path_length",
os.path.join(self.folder, filename),
"Critical Path{} [ms]", "Num. workflows (CDF)", show)
return filename
def generate_graphs(self, show=False):
filename = "task_resource_time_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
self.df = self.df.withColumn("task_resource_computation_time", F.col("resource_amount_requested") * F.col("runtime"))
generate_cdf(self.df, "task_resource_computation_time", os.path.join(self.folder, filename),
"Task Core Time{} [ms]", "Num. tasks [CDF]", show)
return filename
def generate_graphs(self, show=False):
filename = "task_runtime_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
generate_cdf(self.df, "runtime",
os.path.join(self.folder, filename), "Runtime{} [ms]",
"Num. tasks [CDF]", show)
return filename
def generate_graphs(self, show=False):
filename = "job_arrival_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
generate_cdf(self.df, "ts_submit",
os.path.join(self.folder, filename),
"Arrival time{} [ms]", "Num. workflows (CDF)", show)
return filename
def generate_graphs(self, show=False):
filename = "job_cp_task_count_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
generate_cdf(self.df, "critical_path_task_count",
os.path.join(self.folder,
filename), "Num. tasks on critical path",
"Num. workflows (CDF)", show)
return filename
def generate_graphs(self, show=False):
filename = "job_wait_time_cdf_{0}.png".format(self.workload_name)
if not os.path.isfile(os.path.join(self.folder, filename)):
# Add a column so we can compute the CDF based on that
# Null values should be filtered out, but just in case, set ts_start to -2 billion and ts_submit to -1 billion
# if they are null, so they will be filtered out.
self.df.withColumn(
"wf_wait_time",
F.when(F.col("ts_start").isNull(),
F.lit(-2000000000)).otherwise(F.col("ts_start")) -
F.when(F.col("ts_submit").isNull(),
F.lit(-1000000000)).otherwise(F.col("ts_submit")))
generate_cdf(self.df, "wf_wait_time",
os.path.join(self.folder, filename),
"Wait time{} [ms]", "Num. workflows (CDF)", show)
return filename