Luis Tomás

Improving cloud infrastructure utilization through overbooking

Despite the potential given by the combination of multi-tenancy and virtualization, resource utilization in todays data centers is still low. We identify three key characteristics of cloud services and infrastructure as-a-service management practices: burstiness in service workloads, fluctuations in virtual machine resource usage over time, and virtual machines being limited to pre-defined sizes only. Based on these characteristics, we propose scheduling and admission control algorithms that incorporate resource overbooking to improve utilization. A combination of modeling, monitoring, and prediction techniques is used to avoid overpassing the total infrastructure capacity. A performance evaluation using a mixture of workload traces demonstrates the potential for significant improvements in resource utilization while still avoiding overpassing the total capacity.

An Autonomic Approach to Risk-Aware Data Center Overbooking

Elasticity is a key characteristic of cloud computing that increases the flexibility for cloud consumers, allowing them to adapt the amount of physical resources associated to their services over time in an on-demand basis. However, elasticity creates problems for cloud providers as it may lead to poor resource utilization, specially in combination with other factors, such as user overestimations and pre-defined VM sizes. Admission control mechanisms are thus needed to increase the number of services accepted, raising the utilization without affecting services performance. This work focuses on implementing an autonomic risk-aware overbooking architecture capable of increasing the resource utilization of cloud data centers by accepting more virtual machines than physical available resources. Fuzzy logic functions are used to estimate the associated risk to each overbooking decision. By using a distributed PID controller approach, the system is capable of self-adapting over time-changing the acceptable level of risk-depending on the current status of the cloud data center. The suggested approach is extensively evaluated using a combination of simulations and experiments executing real cloud applications with real-life available workloads. Our results show a 50 percent increment at both resource utilization and capacity allocated with acceptable performance degradation and more stable resource utilization over time.

Network-aware meta-scheduling in advance with autonomous self-tuning system

The provision of Quality of Service (QoS) in Grid environments is still an open issue that needs attention from the research community. One way of contributing to the provision of QoS in Grids is by performing meta-scheduling of jobs in advance, that is, jobs are scheduled some time before they are actually executed. In this way, it becomes more likely that the appropriate resources are available to run the job when needed, so that QoS requirements of jobs are met (i.e. jobs are finished within a deadline). This paper presents a framework built on top of Globus and the GridWay meta-scheduler to provide QoS by means of performing meta-scheduling in advance. Thanks to this, QoS requirements of jobs are met. This framework manages idle/busy periods of resources in order to choose the most suitable resource for each job, and uses red-black trees for this task. Besides, no prior knowledge on the duration of jobs is required, as opposed to other works using similar techniques. This framework uses heuristics that consider the network as a first level resource. Furthermore, this framework presents an autonomous behaviour so that it adapts to the dynamic changes of the Grid resources. The autonomous behaviour is obtained by means of computing a trust for each resource and performing job rescheduling. All this set of features make this framework suitable for real Grids. Finally, a performance evaluation using a real testbed is presented that illustrates the efficiency of this approach to meet the QoS requirements of users.

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

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 non-interactive 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.

Exponential Smoothing for Network-Aware Meta-scheduler in Advance in Grids

Grid computing involves the coordinated use of disperse heterogeneous computing resources. This heterogeneity and dispersion makes Quality of Service (QoS) still an open issue requiring attention from the research community. One way of contributing to the provision of QoS in Grids is by performing meta-scheduling of jobs in advance, that is, the computing resource where a job will be executed is decided some time before jobs are actually executed. But this way of scheduling needs to do predictions about the future status of resources, including network. The main aim of this work is to provide QoS in Grid environments through network-aware job scheduling in advance. In our case, QoS means the fulfillment of a deadline for the completion of jobs. For this, predictions about future status of computing and network resources are made by using exponential smoothing functions. This paper presents a performance evaluation using a real testbed that illustrates the efficiency of this approach to meet the QoS requirements of users. This evaluation highlights the effects of using Exponential Smoothing (ES) to tune predictions in order to deliver the requested QoS.

Using network information to perform meta-scheduling in advance in grids

In extremely heterogeneous and distributed systems, like Grid environments, it is quite difficult to provide quality of service (QoS). In addition, the dynamic behaviour of the resources makes the time needed to complete the execution of a job highly variable. So, fulfilling the user QoS requirements in a Grid is still an open issue. The main aim of this work is to provide QoS in Grid environments through network-aware job scheduling in advance. This paper presents a technique to manage idle/busy periods of resources using red-black trees which considers the network as a first level resource. Besides, no a priori knowledge on the duration of jobs is required, as opposed to other works. A performance evaluation using a real testbed is presented which illustrates the efficiency of this approach to meet the QoS requirements of users, and highlights the importance of taking the network into account when predicting the duration of jobs.

Cloudy with a Chance of Load Spikes: Admission Control with Fuzzy Risk Assessments

Elasticity is key for the cloud paradigm, where the pay-per use nature provides great flexibility for end-users. However, elasticity complicates long-term capacity planning for cloud providers as the exact amount of resources required over time becomes uncertain. Admission control techniques are thus needed to handle the trade-off between resource utilization and potential overload. We define a set of admission control algorithms that combine risk assessment methods with a fuzzy aggregation framework. An experimental evaluation using a mixture of bursty and steady applications demonstrate that our algorithms can increase resource utilization by a factor of two while limiting overload problems to a few percent.

A GridWay-based autonomic network-aware metascheduler

One of the key motivations of computational and data grids is the ability to make coordinated use of heterogeneous computing resources which are geographically dispersed. Consequently, the performance of the network linking all the resources present in a grid has a significant impact on the performance of an application. It is therefore essential to consider network characteristics when carrying out tasks such as scheduling, migration or monitoring of jobs. This work focuses on an implementation of an autonomic network-aware meta-scheduling architecture that is capable of adapting its behavior to the current status of the environment, so that jobs can be efficiently mapped to computing resources. The implementation extends the widely used GridWay meta-scheduler and relies on exponential smoothing to predict the execution and transfer times of jobs. An autonomic control loop (which takes account of CPU use and network capability) is used to alter job admission and resource selection criteria to improve overall job completion times and throughput. The implementation has been tested using a real testbed involving heterogeneous computing resources distributed across different national organizations.

Meta-scheduling in advance using red-black trees in heterogeneous Grids

The provision of Quality of Service in Grid environments is still an open issue that needs attention from the research community. One way of contributing to the provision QoS in Grids is by performing meta-scheduling of jobs in advance, that is, jobs are scheduled some time before they are actually executed. In this way, the aproppriate resources will be available to run the job when needed, so that QoS requirements (i.e., deadline) are met. This paper presents two new techniques, implemented over the red-black tree data structure, to manage the idle/busy periods of resources. One of them takes into account the heterogeneity of resources when estimating the execution times of jobs. A performance evaluation using a real testbed is presented that illustrates the efficiency of this approach to meet the QoS requirements of users.

Studying the Influence of Network-Aware Grid Scheduling on the Performance Received by Users

Grid computing is the key enabling technology to aggregate geographically distributed resources in the context of a particular application. As Grids are extremely distributed systems, requirements on the communication network should also be taken into account when performing usual tasks such as scheduling, migrating or monitoring of jobs. Note that users, services, and data need to communicate with each other over networks, thus the network should be used in an efficient and fault-tolerant way. There are Grid schedulers that consider the network when performing their tasks, but the way they have been implemented does not allow easy extensions. Thus, they are not suitable to be modified and try different scheduling approaches. The authors have extended the GridWay metascheduler to perform scheduling considering the network status. This is the first step in order to proceed with more complicated and efficient scheduling and reservation processes. In this work, the extension has been evaluated by means of a testbed, in which users simultaneously submit different jobs with different frequencies to GridWay. Results presented here show that the response time perceived by Grid users is reduced when data on network performance are considered in the job scheduling process.