Transient provisioning and performance evaluation for cloud computing platforms: A capacity value approach

Abstract

User demand on the computational resources of cloud computing platforms varies over time. These variations in demand can be predictable or unpredictable, resulting in 'bursty' fluctuations in demand. Furthermore, demand can arrive in batches, and users whose demands are not met can be impatient. We demonstrate how to compute the expected revenue loss over a finite time horizon in the presence of all these model characteristics through the use of matrix analytic methods. We then illustrate how to use this knowledge to make frequent short term provisioning decisions -transient provisioning. It is seen that taking each of the characteristics of fluctuating user demand (predictable, unpredictable, batchy) into account can result in a substantial reduction of losses. Moreover, our transient provisioning framework allows for a wide variety of system behaviors to be modeled and gives simple expressions for expected revenue loss which are straightforward to evaluate numerically.

An optimization approach to capacity evaluation and investment decision of hybrid cloud: a corporate customer’s perspective

“…For example, a cloud center is modeled as a classic open network with a single arrival from which the distribution of response time is obtained, assuming that both interarrival and service times are exponential [17]. A recent study of Patch and Taimre [18] also assumes that tasks require an exponentially distributed service time for transient provisioning and performance evaluation of cloud computing platforms.…”

“…They analyze the cost side aspect with a simulation model that not only helps assess the suitability of the cloud computing but also measures its profitability. Recently, Patch and Taimre [18] demonstrate that taking into account each of the characteristics of fluctuating user demand of cloud services can result in a substantial reduction of losses. Their transient provisioning framework allows for a wide variety of system behaviors to be modeled in the presence of various model characteristics through the use of matrix analytic methods to minimize the revenue loss over a finite time horizon.…”

Optimising capacity allocation in networks of stochastic loss systems: A functional-form approach

“…where • denotes the usual Euclidean inner product. This expression serves as our objective function, which is inspired by the capacity value function investigated by [10,9] in the context of Erlang-B loss systems and later generalised by [26] in the context of provisioning cloud computing platforms. Our optimisation formulation seeks to determine…”

“…Evaluation of the objective function (1) given in Section 2 for a single station one class Markovian system is quite straightforward on a modern computer, for more complex instances this is not the case. The power of the approach described in [26], based on matrix analytic methods (MAMs), is that more general systems can be considered, such as the example given in Section 2. This section uses these methods to explicitly compute our objective for this example, so that it can be used to exactly test our simulation method.…”

“…Let X 1,1 (t) and X 1,2 (t) be the number of customers being processed by the system at the first and second stations respectively. Following [26], we observe that this description leads to a marked MAP (see, e.g. [15, Section 2.5] for details), N 1 (t), N 2 (t), t ∈ R 0 , that counts the number of customers successfully entering the first and second stations as follows.…”