Xenia Mountrouidou

Learn More
We characterize analytically the departure process from the following three burst ag-gregation algorithms: the time based aggregation algorithm, the burst-length based aggregation algorithm and the time and burst-length based aggregation algorithm. The arrival process of packets is assumed to be Poisson or bursty modeled by an Interrupted Poisson Process(More)
We describe an analytic approach for the calculation of the departure process from a burst ggregation algorithm that uses both a timer and maximum/minimum burst size. The arrival process of packets is assumed to be Poisson or bursty modelled by an Interrupted Pois-son Process (IPP). The analytic results are approximate and validation against simulation data(More)
We describe an efficient and accurate approximation method for calculating the bandwidth that should be allocated on each link along the path of a point-to-point MPLS connection, so that the end-to-end delay D is less than or equal to a given value T with a probability γ, that is, P(D≤T) = γ. We model a connection by a tandem queueing network of infinite(More)
We obtain analytically the delay per packet in a burst for the three burst aggregation algorithms: Time based, Burst Length based and mixed Time and Burst Length based, assuming real video and aggregated Internet traces, as well as the IPP arrivals. For video traces we also obtain the probability that a playback machine will pause due to lack of frames when(More)
In order to reduce the amount of power consumption in data centers, it is becoming necessary to shut off or slow down disks that are not actively serving user requests. In addition to exploiting disk drive idleness, system features are in place that shape a disk's workload by redirecting portions of it elsewhere, with the goal to expand the periods of(More)
The biggest power consumer in data centers is the storage system. Coupled with the fact that disk drives are lowly utilized, disks offer great opportunities for power savings, but any power saving action should be transparent to user traffic. Estimating correctly the performance impact of power saving becomes crucial for the effectiveness of power saving.(More)
—With most of today's systems being highly distributed , from data centers to cloud and storage clusters, there is a prevalent need for robust methodologies for work consolidation to improve load balancing but also to optimize non-traditional performance measures. Such alternative measures may include power savings, e.g., it may be desirable to shut down a(More)
As storage in data centers is increasing rapidly, it has become critical to find ways to operate efficiently this important component of a data center. Often, it has been proposed to consolidate the storage workload into a subset of storage devices and shutdown the unused ones with the purpose of preserving power. In many cases storage workload(More)
Performance studies point to the fact that in an OBS network , the link utilization has to be kept very low in order for the burst loss probability to be within an acceptable level. Various congestion control schemes have been proposed, such as the use of converters, fiber delay lines, and deflection routing. However, these schemes do not alleviate this(More)
—We present a robust framework that aims at harvesting future idle intervals for power savings within strict constraints: first, it is imperative to contain the delays in service of IO requests that occur during power savings since the time to bring up the disk is not negligible and second, ensure that the power saving mechanism is triggered few times only,(More)