The Straw that Broke the Camel's Back: Safe Cloud Overbooking with Application Brownout

Abstract

Abstract-Resource overbooking is an admission control technique to increase utilization in cloud environments. However, due to uncertainty about future application workloads, overbooking may result in overload situations and deteriorated performance. We mitigate this using brownout, a feedback approach to application performance steering, that ensures graceful degradation during load spikes and thus avoids overload. Additionally, brownout management information is included into the overbooking system, enabling the development of improved reactive methods to overload situations. Our combined brownout-overbooking approach is evaluated based on real-life interactive workloads and noninteractive batch applications. The results show that our approach achieves an improvement of resource utilization of 11 to 37 percentage points, while keeping response times lower than the set target of 1 second, with negligible application degradation.

Event-Driven Application Brownout: Reconciling High Utilization and Low Tail Response Times

“…This effect could alter results obtained with the load balancing algorithms, therefore it is worth to investigate the event-driven Brownout action with such algorithms. In addition, the initial periodic Brownout has been included within an overbooking system, where the purpose was to develop and improve reactive methods in case of overload situations [34]. Again it would be worth investigating the effect of the event-driven Brownout on this work, to see whether the results differ.…”

Reducing Noisy-Neighbor Impact with a Fuzzy Affinity-Aware Scheduler

“…However, once the applications are admitted, they do not provide any mechanism to deal with resource shortages due to mispredictions or incorrect overbooking decisions, which may lead to performance degradation, specially for latency-critical applications. To tackle this problem we allowed our framework to adjust the target utilization levels (and consequently the overbooking pressure) based on applications performance input, while at the same time, some applications can use the brownout approach [14] to mitigate short-term problems until the system adapts to the new situation [15].…”

Incentivizing self-capping to increase cloud utilization

“…With proper coordination between an SP and its IP, brownout can be used to compensate for overbooking [63]. In this work, we tackle the real deployment of such methods by incentivizing tenants to adopt them.…”

“…At the infrastructure level, mapping algorithms are used to co-locate applications with antagonist resource requirements [13,43,66]. Since most IaaS tenants tend to overprovision their VM demands, resource overbooking (over-commit) [23,63,64] was used to accept more tenants if the resource usage is predicted to allow so without SLA violations. Performance-based service diferentiation [36] may be used to ensure łgoldž applications always maintain their target performance, while łbronzež ones are degraded if actual usage was mispredicted.…”

“…Self-adaptation is also applied to cloud applications allowing them to survive temporary capacity shortages by degrading or disabling some of their features [33]. Researchers have already shown that using graceful degradation can improve the cloud resource utilization by 11 to 37 percentage points [63]. Moreover, it is easier to meet latency requirements for less bursty tenants [78].…”