Delay and Price Differentiation in Cloud Computing: A Service Model, Supporting Architectures, and Performance

Many cloud service providers (CSPs) offer an on-demand with a small delay. Motivated by the reality of ecosystems, we study non-interruptible services and consider differentiated model to complement existing market offering multiple level agreements (SLAs) satisfy users different delay tolerance. The itself is incentive compatible construction. Two typical architectures are considered fulfill SLAs: (i) non-preemptive priority queues (ii) independent groups servers. We leverage queueing theory establish guidelines for resultant market: (a) Under first architecture, can only improve revenue marginally over pure (b) under second give closed-form expression improvement when CSP offers two SLAs derive condition which viable. Additionally, exhaustive search procedure find optimal SLA delays prices generally SLAs. Numerical results show that achieved be significant even if offered. Our help CSPs design delay-differentiated choose appropriate serving architectures.