Combining Futures and Spot Markets: A Hybrid Market Approach to Economic Grid Resource Management

Abstract

Economic forms of resource management in which users can express their valuations for service, offer new possibilities for optimizing resource allocations in Grids. If users are to correctly express these valuations, quality of service guarantees need to be given with respect to the turnaround time of their workloads. Market mechanisms that support bidding and allocations in future time are crucial for delivering such guarantees. To deal with the significant delays that these mechanisms introduce in the allocation process, we present a hybrid market approach in which a low-latency spot market coexists with a higher latency futures market. Based on simulated market scenarios, we show how this combination can significantly increase the total value realized by the Grid infrastructure. We also demonstrate how providers can react to price dynamics in such a hybrid market setting.

IT controls in the public cloud: Success factors for allocation of roles and responsibilities

“…The management of computational resources in an organisation is broadly characterised by the cloud deployment model adopted by the organisation (Jansen & Grance, 2011). A variety of deployment models of cloud services have evolved in the marketplace (Krieger, McGachey, & Kanevsky, 2010;Vanmechelen, Depoorter, & Broeckhove, 2011;Venkatraman, 2013), with public, private, hybrid and community cloud deployment models being the most popular ones. While a private cloud is a shared multi-tenant environment built on a highly efficient automated and virtualised infrastructure using in-house resources and can belong to an organisation solely (Pathak et al, 2012), a public cloud represents a publicly accessible distributed system hosting the execution of applications and providing services billed on a pay-per-use basis (Mattess, Vecchiola, Garg, & Buyya, 2011).…”

A coordinator for scaling elastic applications across multiple clouds

“…Works related to systems for market-making such as GridEcon [18], SORMA [19], and Mandi [5], as well as the works by Song et al [20], Mihailescu and Teo [21], Gomes et al [22], and Vanmechelen et al [23] concern mechanisms for creating markets, trading resources (e.g., auctions or fixed price), and also mechanisms to motivate market maintenance, i.e., study and development of techniques that motivate both resource providers and resource consumers to join and stay in the market.…”

“…Alternatively, the Cloud Exchange can host auctions for resources by leveraging the Mandi architecture [5]. The presence of a market-maker component allows static pricing policies (as currently adopted by most of the Cloud providers) to be replaced by dynamic pricing policies, which can increase profit of providers [31,23,22].…”

Characterizing spot price dynamics in public cloud environments

“…checkpointing and replication algorithms) for spot market in public Cloud environments. In addition, although Amazon is the only provider of SIs at the moment, some researches have been conducted to analyze the free computing resource markets [6,7]. So, this model can be used by other resource providers that look forward to offer such a service in the near future.…”