Cataclysm: Scalable overload policing for internet applications

Abstract

In this paper we present the Cataclysm server platform for handling extreme overloads in hosted Internet applications. The primary contribution of our work is to develop a low overhead, highly scalable admission control technique for Internet applications. Cataclysm provides several desirable features, such as guarantees on response time by conducting accurate size-based admission control, revenue maximization at multiple time-scales via preferential admission of important requests and dynamic capacity provisioning, and the ability to be operational even under extreme overloads. Cataclysm can transparently trade-off the accuracy of its decision making with the intensity of the workload allowing it to handle incoming rates of several tens of thousands of requests/second. We implement a prototype Cataclysm hosting platform on a Linux cluster and demonstrate the benefits of our integrated approach using a variety of workloads. r

SOFTScale: Stealing Opportunistically for Transient Scaling

“…For multi-tiered cloud computing environments, SOFTScale can be used in conjunction with statistical multiplexing. Finally, there is also a lot of prior work [37][38][39][40] that deals with managing overload conditions by allowing for performance degradation. Some of the popular techniques that have been used to regulate performance degradation include admission control and request prioritization.…”

“…By contrast, SOFTScale handles load spikes without any performance degradation, provided the load is not high. If the load is high, SOFTScale can be coupled with techniques like those in [37][38][39][40] to minimize the damage caused by load spikes.…”

Optimal resource allocation in synchronized multi-tier Internet services

“…Villela et al [20] use an M/G/1/PS queueing model for business logic tiers and to provision the capacity of application servers. A G/G/1 based queueing model for modeling replicated Web servers is proposed in [19], which is to perform admission control during overloads. In contrast to these queueing based approaches, a feedback control based model was proposed in [1].…”

Intelligent Overload Control for Composite Web Services

“…Now dynamic admission control algorithm S is derived based on the following constraint: the expected sojourn time E(S) of a job in a web service should be less than or equal to the average available time. Now our optimization problem is defined as follows: (4) In (4), both c and R are time-dependent, but we omit this to simplify our notation. Computation of R is straightforward since due times of all jobs within the composite service are known.…”

“…Research has also been performed on the use of admission control for Web Servers; see for instance [1]- [3], [7]. The use of WAC to prevent overload for stand alone Web Services has been discussed in [4]- [5]. In the field of composite web services several contributions have been made more recently, focusing on web service scheduling; for instance in [8]- [9].…”