codex-storage
/
das-research
mirror of https://github.com/codex-storage/das-research.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'develop' into addDiagnostics

This commit is contained in:
Leonardo Bautista-Gomez 2023-03-30 13:56:09 +02:00
parent 4b7bf81cee b5390b9f1b
commit 699a912991
6 changed files with 39 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
DAS/observer.py
View File

@ -58,6 +58,17 @@ class Observer:
return (arrived, expected, ready, validated)
def getProgress(self, validators):
"""Calculate current simulation progress with different metrics.
Returns:
- missingSamples: overall number of sample instances missing in nodes.
Sample are counted on both rows and columns, so intersections of interest are counted twice.
- sampleProgress: previous expressed as progress ratio
- nodeProgress: ratio of nodes having all segments interested in
- validatorProgress: same as above, but vpn weighted average. I.e. it counts per validator,
but counts a validator only if its support node's all validators see all interesting segments
TODO: add real per validator progress counter
"""
arrived, expected, ready, validated = self.checkStatus(validators)
missingSamples = expected - arrived
sampleProgress = arrived / expected
@ -68,6 +79,7 @@ class Observer:
return missingSamples, sampleProgress, nodeProgress, validatorProgress
def getTrafficStats(self, validators):
"""Summary statistics of traffic measurements in a timestep."""
def maxOrNan(l):
return np.max(l) if l else np.NaN
def meanOrNan(l):

16
DAS/results.py
View File

@ -7,35 +7,37 @@ from dicttoxml import dicttoxml
class Result:
"""This class stores and process/store the results of a simulation."""
def __init__(self, shape):
def __init__(self, shape, execID):
"""It initializes the instance with a specific shape."""
self.shape = shape
self.execID = execID
self.blockAvailable = -1
self.tta = -1
self.missingVector = []
self.metrics = {}
def populate(self, shape, missingVector):
def populate(self, shape, config, missingVector):
"""It populates part of the result data inside a vector."""
self.shape = shape
self.missingVector = missingVector
missingSamples = missingVector[-1]
if missingSamples == 0:
self.blockAvailable = 1
self.tta = len(missingVector)
self.tta = len(missingVector) * (1000/config.stepDuration)
else:
self.blockAvailable = 0
self.tta = -1
def addMetric(self, name, metric):
"""Generic function to add a metric to the results."""
self.metrics[name] = metric
def dump(self, execID):
def dump(self):
"""It dumps the results of the simulation in an XML file."""
if not os.path.exists("results"):
os.makedirs("results")
if not os.path.exists("results/"+execID):
os.makedirs("results/"+execID)
if not os.path.exists("results/"+self.execID):
os.makedirs("results/"+self.execID)
resd1 = self.shape.__dict__
resd2 = self.__dict__.copy()
resd2.pop("shape")
@ -43,6 +45,6 @@ class Result:
resXml = dicttoxml(resd1)
xmlstr = minidom.parseString(resXml)
xmlPretty = xmlstr.toprettyxml()
filePath = "results/"+execID+"/"+str(self.shape)+".xml"
filePath = "results/"+self.execID+"/"+str(self.shape)+".xml"
with open(filePath, "w") as f:
f.write(xmlPretty)

18
DAS/simulator.py
View File

@ -19,7 +19,8 @@ class Simulator:
self.shape = shape
self.config = config
self.format = {"entity": "Simulator"}
self.result = Result(self.shape)
self.execID = execID
self.result = Result(self.shape, self.execID)
self.validators = []
self.logger = []
self.logLevel = config.logLevel
@ -279,7 +280,7 @@ class Simulator:
missingVector.append(missingSamples)
break
elif missingSamples == 0:
#self.logger.info("The entire block is available at step %d, with failure rate %d !" % (steps, self.shape.failureRate), extra=self.format)
self.logger.debug("The entire block is available at step %d, with failure rate %d !" % (steps, self.shape.failureRate), extra=self.format)
missingVector.append(missingSamples)
break
else:
@ -288,17 +289,6 @@ class Simulator:
progress = pd.DataFrame(progressVector)
if self.config.saveProgress:
self.result.addMetric("progress", progress.to_dict(orient='list'))
if self.config.plotProgress:
progress.plot.line(subplots = [[cnS, cnN, cnV], [cnT0], [cnT1, cnR1, cnD1], [cnT2, cnR2, cnD2]],
title = str(self.shape))
if not os.path.exists("results"):
os.makedirs("results")
if not os.path.exists("results/"+self.execID):
os.makedirs("results/"+self.execID)
filePath = "results/"+self.execID+"/"+str(self.shape)+".png"
matplotlib.pyplot.savefig(filePath)
matplotlib.pyplot.close()
self.result.populate(self.shape, missingVector)
self.result.populate(self.shape, self.config, missingVector)
return self.result

6
DAS/visualizer.py
View File

@ -36,7 +36,7 @@ class Visualizer:
bwUplinkProd = int(root.find('bwUplinkProd').text)
bwUplink1 = int(root.find('bwUplink1').text)
bwUplink2 = int(root.find('bwUplink2').text)
tta = int(root.find('tta').text)
tta = float(root.find('tta').text)
# Loop over all possible combinations of of the parameters minus two
for combination in combinations(self.parameters, len(self.parameters)-2):
@ -120,7 +120,7 @@ class Visualizer:
hist, xedges, yedges = np.histogram2d(data[key][labels[0]], data[key][labels[1]], bins=(len(xlabels), len(ylabels)), weights=data[key]['ttas'])
hist = hist.T
fig, ax = plt.subplots(figsize=(10, 6))
sns.heatmap(hist, xticklabels=xlabels, yticklabels=ylabels, cmap='Purples', cbar_kws={'label': 'Time to block availability'}, linecolor='black', linewidths=0.3, annot=True, fmt=".2f", ax=ax)
sns.heatmap(hist, xticklabels=xlabels, yticklabels=ylabels, cmap='Purples', cbar_kws={'label': 'Time to block availability (ms)'}, linecolor='black', linewidths=0.3, annot=True, fmt=".2f", ax=ax)
plt.xlabel(self.formatLabel(labels[0]))
plt.ylabel(self.formatLabel(labels[1]))
filename = ""
@ -131,6 +131,8 @@ class Visualizer:
filename += f"{key[paramValueCnt]}"
formattedTitle = self.formatTitle(key[paramValueCnt])
title += formattedTitle
if (paramValueCnt+1) % 5 == 0:
title += "\n"
paramValueCnt += 1
title_obj = plt.title(title)
font_size = 16 * fig.get_size_inches()[0] / 10

15
config_example.py
View File

@ -31,14 +31,14 @@ logLevel = logging.INFO
# number of parallel workers. -1: all cores; 1: sequential
# for more details, see joblib.Parallel
numJobs = 3
numJobs = -1
# distribute rows/columns evenly between validators (True)
# or generate it using local randomness (False)
evenLineDistribution = True
# Number of simulation runs with the same parameters for statistical relevance
runs = range(10)
runs = range(2)
# Number of validators
numberNodes = range(256, 513, 128)
@ -53,14 +53,14 @@ blockSizes = range(32,65,16)
netDegrees = range(6, 9, 2)
# number of rows and columns a validator is interested in
chis = range(1, 5, 2)
chis = range(2, 5, 2)
# ratio of class1 nodes (see below for parameters per class)
class1ratios = np.arange(0, 1, .2)
class1ratios = [0.8, 0.9]
# Number of validators per beacon node
validatorsPerNode1 = [1]
validatorsPerNode2 = [2, 4, 8, 16, 32]
validatorsPerNode2 = [500]
# Set uplink bandwidth. In segments (~560 bytes) per timestep (50ms?)
# 1 Mbps ~= 1e6 / 20 / 8 / 560 ~= 11
@ -68,8 +68,11 @@ bwUplinksProd = [2200]
bwUplinks1 = [110]
bwUplinks2 = [2200]
# Step duration in miliseconds (Classic RTT is about 100ms)
stepDuration = 50
# Set to True if you want your run to be deterministic, False if not
deterministic = False
deterministic = True
# If your run is deterministic you can decide the random seed. This is ignore otherwise.
randomSeed = "DAS"

2
study.py
View File

@ -36,7 +36,7 @@ def runOnce(config, shape, execID):
sim.logger.info("Shape: %s ... Block Available: %d in %d steps" % (str(sim.shape.__dict__), result.blockAvailable, len(result.missingVector)), extra=sim.format)
if config.dumpXML:
result.dump(execID)
result.dump()
return result