eugen.betke 2020-08-20 14:17:03 +02:00
commit 75fe1a9952
8 changed files with 130 additions and 33 deletions

Binary file not shown.

Binary file not shown.

Binary file not shown.

Binary file not shown.

View File

@ -44,7 +44,8 @@
\usepackage{graphicx}
\graphicspath{
{./pictures/}
{./pictures/},
{../fig/}
}
\usepackage[backend=bibtex, style=numeric]{biblatex}
@ -127,30 +128,62 @@ Check time series algorithms:
\begin{itemize}
\item bin
\item hex\_native/hex\_lev
\item pm\_quant
\item hex\_native
\item hex\_lev
\item hex\_quant
\end{itemize}
\section{Evaluation}
\label{sec:evaluation}
Two study examples (two reference jobs):
In the following, we assume a job is given and we aim to identify similar jobs.
We chose several reference jobs with different compute and IO characteristics visualized in \Cref{fig:refJobs}:
\begin{itemize}
\item jobA: shorter length, e.g. 5-10, that has a little bit IO in at least two metadata metrics (more better).
\item jobB: a very IO intensive longer job, e.g., length $>$ 20, with IO read or write and maybe one other metrics.
\item Job-S: performs postprocessing on a single node. This is a typical process in climate science where data products are reformatted and annotated with metadata to a standard representation (so called CMORization). The post-processing is IO intensive.
\item Job-M: a typical MPI parallel 8-hour compute job on 128 nodes which writes time series data after some spin up. %CHE.ws12
\item Job-L: a 66-hour 20-node job.
The initialization data is read at the beginning.
Then only a single master node writes constantly a small volume of data; in fact, the generated data is too small to be categorized as IO relevant.
\end{itemize}
For each reference job: create CSV file which contains all jobs with:
\begin{itemize}
\item JOB ID, for each algorithm: the coding and the computed ranking $\rightarrow$ thus one long row.
\end{itemize}
Alternatively, could be one CSV for each algorithm that contains JOB ID, coding + rank
For each reference job and algorithm, we created a CSV files with the computed similarity for all other jobs.
Sollte man was zur Laufzeit der Algorithmen sagen? Denke Daten zu haben wäre sinnvoll.
Create histograms + cumulative job distribution for all algorithms.
Insert job profiles for closest 10 jobs.
Potentially, analyze how the rankings of different similarities look like.
\begin{figure}
\begin{subfigure}{0.8\textwidth}
\includegraphics[width=\textwidth]{job-timeseries4296426}
\caption{Job-S} \label{fig:job-S}
\end{subfigure}
\caption{Reference jobs: timeline of mean IO activity}
\label{fig:refJobs}
\end{figure}
\begin{figure}\ContinuedFloat
\begin{subfigure}{0.8\textwidth}
\includegraphics[width=\textwidth]{job-timeseries5024292}
\caption{Job-M} \label{fig:job-M}
\end{subfigure}
\begin{subfigure}{0.8\textwidth}
\includegraphics[width=\textwidth]{job-timeseries7488914-30.pdf}
\caption{Job-L (first 30 segments of 400; remaining segments are similar)}
\label{fig:job-L}
\end{subfigure}
\caption{Reference jobs: timeline of mean IO activity; non-shown timelines are 0}
\end{figure}
\section{Summary and Conclusion}
\label{sec:summary}

View File

@ -0,0 +1,14 @@
#!/bin/bash
# This script calls all other scripts to re-create the figures for the paper
mkdir fig
for job in 5024292 4296426 7488914 ; do
./scripts/plot-single-job.py $job "fig/job-"
done
# Remove whitespace around jobs
# for file in fig/*.pdf ; do
# pdfcrop $file output.pdf
# mv output.pdf $file
# done

View File

@ -5,12 +5,47 @@ import sys
from pandas import DataFrame
from pandas import Grouper
from matplotlib import pyplot
import matplotlib.cm as cm
jobs = [sys.argv[1]]
prefix = sys.argv[2]
jobs = sys.argv[1].split(",")
prefix = sys.argv[2].split(",")
print("Plotting the job: " + str(jobs))
# Color map
colorMap = { "md_file_create": cm.tab10(0),
"md_file_delete": cm.tab10(1),
"md_mod": cm.tab10(2),
"md_other": cm.tab10(3),
"md_read": cm.tab10(4),
"read_bytes": cm.tab10(5),
"read_calls": cm.tab10(6),
"write_bytes": cm.tab10(7),
"write_calls": cm.tab10(8)
}
markerMap = { "md_file_create": "^",
"md_file_delete": "v",
"md_other": ".",
"md_mod": "<",
"md_read": ">",
"read_bytes": "h",
"read_calls": "H",
"write_bytes": "D",
"write_calls": "d"
}
linestyleMap = { "md_file_create": ":",
"md_file_delete": ":",
"md_mod": ":",
"md_other": ":",
"md_read": ":",
"read_bytes": "--",
"read_calls": "--",
"write_bytes": "-.",
"write_calls": "-."
}
# Plot the timeseries
def plot(prefix, header, row):
x = { h : d for (h, d) in zip(header, row)}
@ -36,27 +71,45 @@ def plot(prefix, header, row):
groups = data.groupby(["metrics"])
metrics = DataFrame()
labels = []
colors = []
style = []
for name, group in groups:
metrics[name] = [x[2] for x in group.values]
labels.append(name)
style.append(linestyleMap[name] + markerMap[name])
colors.append(colorMap[name])
ax = metrics.plot(subplots=True, legend=False, sharex=True, grid = True, sharey=True, colormap='jet', marker='.', markersize=10, figsize=(8, 2 + 2 * len(labels)))
for (i, l) in zip(range(0, len(labels)), labels):
ax[i].set_ylabel(l)
fsize = (8, 1 + 1.5 * len(labels))
fsizeFixed = (8, 2)
pyplot.close('all')
if len(labels) < 4 :
ax = metrics.plot(legend=True, sharex=True, grid = True, sharey=True, markersize=10, figsize=fsizeFixed, color=colors, style=style)
ax.set_ylabel("Value")
else:
ax = metrics.plot(subplots=True, legend=False, sharex=True, grid = True, sharey=True, markersize=10, figsize=fsize, color=colors, style=style)
for (i, l) in zip(range(0, len(labels)), labels):
ax[i].set_ylabel(l)
pyplot.xlabel("Segment number")
pyplot.savefig(prefix + "timeseries" + jobid + ".png")
pyplot.savefig(prefix + "timeseries" + jobid + ".pdf", bbox_inches='tight')
# Plot first 30 segments
if len(timeseries) <= 50:
return
ax = metrics.plot(subplots=True, legend=False, sharex=True, grid = True, sharey=True, colormap='jet', marker='.', markersize=10, xlim=(0,30))
for (i, l) in zip(range(0, len(labels)), labels):
ax[i].set_ylabel(l)
if len(labels) < 4 :
ax = metrics.plot(legend=True, xlim=(0,30), sharex=True, grid = True, sharey=True, markersize=10, figsize=fsizeFixed, color=colors, style=style)
ax.set_ylabel("Value")
else:
ax = metrics.plot(subplots=True, xlim=(0,30), legend=False, sharex=True, grid = True, sharey=True, markersize=10, figsize=fsize, color=colors, style=style)
for (i, l) in zip(range(0, len(labels)), labels):
ax[i].set_ylabel(l)
pyplot.xlabel("Segment number")
pyplot.savefig(prefix + "timeseries" + jobid + "-30.png")
pyplot.savefig(prefix + "timeseries" + jobid + "-30.pdf", bbox_inches='tight')
### end plotting function
@ -65,6 +118,7 @@ def plot(prefix, header, row):
with open('job-io-datasets/datasets/job_codings.csv') as csv_file:
csv_reader = csv.reader(csv_file, delimiter=',')
line_count = 0
job = 0
for row in csv_reader:
if line_count == 0:
header = row
@ -74,4 +128,5 @@ with open('job-io-datasets/datasets/job_codings.csv') as csv_file:
if not row[0].strip() in jobs:
continue
else:
plot(prefix, header, row)
plot(prefix[job], header, row)
job += 1

View File

@ -19,10 +19,8 @@ data = read.csv(file)
# Columns are: jobid alg_id alg_name similarity
data$alg_id = as.factor(data$alg_id)
print(nrow(data))
# FILTER, TODO
data = data %>% filter(similarity <= 1.0)
cat("Job count:")
cat(nrow(data))
# empirical cummulative density function (ECDF)
ggplot(data, aes(similarity, color=alg_name, group=alg_name)) + stat_ecdf(geom = "step") + xlab("SIM") + ylab("Fraction of jobs") + theme(legend.position="bottom") + scale_color_brewer(palette = "Set2")
@ -34,7 +32,7 @@ print(summary(e))
ggsave("ecdf-0.5.png")
# histogram for the jobs
ggplot(data, aes(similarity), group=alg_name) + geom_histogram(color="black", binwidth=0.025) + aes(fill = alg_name) + facet_grid(alg_name ~ ., switch = 'y') + scale_y_continuous(limits=c(0, 100), oob=squish) + scale_color_brewer(palette = "Set2") + ylab("Count (cropped at 100)")
ggplot(data, aes(similarity), group=alg_name) + geom_histogram(color="black", binwidth=0.025) + aes(fill = alg_name) + facet_grid(alg_name ~ ., switch = 'y') + scale_y_continuous(limits=c(0, 100), oob=squish) + scale_color_brewer(palette = "Set2") + ylab("Count (cropped at 100)") + theme(legend.position = "none")
ggsave("hist-sim.png")
# load job information, i.e., the time series per job
@ -51,13 +49,10 @@ plotJobs = function(jobs){
md = metadata[metadata$jobid %in% jobs,]
print(summary(md))
# print the job timeline
# print the job timelines
r = e[ordered, ]
for (row in 1:length(jobs)) {
prefix = sprintf("%s-%f-%.0f-", level, r[row, "similarity"], row)
job = r[row, "jobid"]
system(sprintf("scripts/plot-single-job.py %s %s", job, prefix))
}
prefix = do.call("sprintf", list("%s-%.0f-", level, r$similarity))
system(sprintf("scripts/plot-single-job.py %s %s", paste(r$jobid, collapse=","), paste(prefix, collapse=",")))
}
# Store the job ids in a table, each column is one algorithm