README.md

TopoMAD

Datasets

MBD

MBD (1).csv : metrics of MBD

MBD_topology.pk: topology of MBD reprensented as an edge array

MMS

MMS (1).csv : metrics of MMS

MMS_topology.pk : topology of MMS reprensented as an edge array

How to Open the Datasets

MBD

DATASET = "../DatasetUpdate/MBD (1).csv"
TOPOLOGY = "../DatasetUpdate/MBD_topology.pk"

import pandas as pd
import numpy as np
import pickle

data = pd.read_csv(DATASET, header=[0,1])

# preprocess
metric = data.drop(['date', 'label'], axis = 1)
metric.columns.names = ['host','metric']
tempm = metric.swaplevel('metric','host',axis=1).stack()
tempm = (tempm-tempm.mean())/(tempm.std())
metric = tempm.unstack().swaplevel('metric','host',axis=1).stack().unstack()

with open(TOPOLOGY, 'rb') as f:
    edge_index = pickle.load(f)

edge_index represents graph connectivity in COO format with shape [2, num_edges]. In edge_index, each node in the topology is represented with its corresponding index. Specifically, the index $i$ corresponds to the node named as the $i^{th}$ object in metric.columns.levels[0]

MMS

DATASET = "../DatasetUpdate/MMS (1).csv"
TOPOLOGY = "../DatasetUpdate/MMS_topology.pk"

import pandas as pd
import numpy as np
import pickle

data = pd.read_csv(DATASET, header=[0,1])

# preprocess
metric = data.drop(['TimeStamp', 'label'], axis = 1)
metric.columns.names = ['pod','metric']
tempm = metric.swaplevel('metric','pod',axis=1).stack()
tempm = (tempm-tempm.mean())/(tempm.std())
metric = tempm.unstack().swaplevel('metric','pod',axis=1).stack().unstack()

with open(TOPOLOGY, 'rb') as f:
    edge_index = pickle.load(f)

edge_index represents graph connectivity in COO format with shape [2, num_edges]. In edge_index, each node in the topology is represented with its corresponding index. Specifically, the index $i$ corresponds to the node named as the $i^{th}$ object in metric.columns.levels[0]