Obscured by the cloud: A resource allocation framework to model cloud outage events

Abstract

As Small Medium Enterprises (SMEs) adopt Cloud technologies to provide high value customer offerings, uptime is considered important. Cloud outages represent a challenge to SMEs and micro teams to maintain a services platform. If a Cloud platform suffers from downtime this can have a negative effect on business revenue. Additionally, outages can divert resources from product development/delivery tasks to reactive remediation. These challenges are immediate for SMEs or micro teams with a small levels of resources. In this paper we present a framework that can model the arrival of Cloud outage events. This framework can be used by DevOps teams to manage their scarce pool of resources to resolve outages, thereby minimising impact to service delivery. We analysed over 300 Cloud outage events from an enterprise data set. We modelled the inter-arrival and service times of each outage event and found a Pareto and a lognormal distribution to be a suitable fit. We used this result to produce a special case of the G/G/1 queue system to predict busy times of DevOps personnel. We also investigated dependence between overlapping outage events. Our predictive queuing model compared favourably with observed data, 72% precision was achieved using one million simulations.

Service Level Agreement Violations in Cloud Storage: Insurance and Compensation Sustainability

“…The model has been proposed by Dunne and Malone [14], who analysed 300+ cloud outage events recorded for an enterprise cloud system (not better identified, but belonging to a small/medium enterprise) over an eighteen-month period. They tested seven models (Pareto, loglogistic, lognormal, gamma, exponential, logistic, and Weibull) through the Anderson-Darling test.…”

“…In that paper, the correct premium has been computed for the case where the outage phenomenon is described by a Poisson process for occurrence of outages and by the generalized Pareto distribution for their duration. However, an alternative model has been recently introduced in the literature, based on an extensive measurement campaign on a private cloud [14] belonging to a small/medium company. In that paper, the interarrival time between successive outages is modelled through a Pareto distribution, while the duration of an outage is modelled by a lognormal distribution.…”

“…In this paper, we again consider the problem of correctly determining the insurance premium against cloud outages. We adopt the expected utility paradigm as in [13], but we consider the outage model recently proposed in [14] in addition to the exponential-Pareto (a.k.a. Poisson-Pareto) examined in [13].…”

Whose Fault is It? Correctly Attributing Outages in Cloud Services

“…The ending time of the outage may be also reported wrongly, since most users do not take on themselves to report it, and we have to rely on the cloud provider announcing that the problem has been solved and the cloud is back to its fully operational state. Statistics of working periods and outages have been modelled in [19] with data coming from a small private cloud. Active measurement systems based on ICMP probing packets have been investigated in [20]- [22].…”

Pricing Cat Bonds for Cloud Service Failures

“…Exponential-exponential model (or Poisson-exponential), employed inMastroeni and Naldi (2011); • Exponential-Pareto (or Poisson-Pareto) that was proposed by Mastroeni et al (2019); Mastroeni and Naldi (2011) based on a dataset of customer-reported outages for five major cloud providers (Google, Amazon, Rackspace, Salesforce, Windows Azure); • Pareto-LogNormal model, proposed byDunne and Malone (2017) to describe the results of a measurement campaign in a small company running its own cloud.…”