Cerberus

Griner, Chen; Zerwas, Johannes; Blenk, Andreas; Ghobadi, Manya; Schmid, Stefan; Avin, Chen

doi:10.1145/3491050

Cited by 28 publications

(28 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Uncertain Resource Demands: The resource demand changes over time in an hour, day, week, month and years with respect to the type of workload and deadline of execution submitted by the user. Sometimes the traffic becomes bursty [24] which makes it difficult to estimate the upcoming resource demands and decide resource distribution.…”

Section: Research Challengesmentioning

confidence: 99%

Performance Analysis of Machine Learning Centered Workload Prediction Models for Cloud

Saxena,

Kumar,

Singh

et al. 2023

Preprint

View full text Add to dashboard Cite

The precise estimation of resource usage is a complex and challenging issue due to the high variability and dimensionality of heterogeneous service types and dynamic workloads. Over the last few years, the prediction of resource usage and traffic has received ample attention from the research community. Many machine learning-based workload forecasting models have been developed by exploiting their computational power and learning capabilities. This paper presents the first systematic survey cum performance analysis-based comparative study of diversified machine learning-driven cloud workload prediction models. The discussion initiates with the significance of predictive resource management followed by a schematic description, operational design, motivation, and challenges concerning these workload prediction models. Classification and taxonomy of different prediction approaches into five distinct categories are presented focusing on the theoretical concepts and mathematical functioning of the existing state-of-the-art workload prediction methods. The most prominent prediction approaches belonging to a distinct class of machine learning models are thoroughly surveyed and compared. All five classified machine learning-based workload prediction models are implemented on a common platform for systematic investigation and comparison using three distinct benchmark cloud workload traces via experimental analysis. The essential key performance indicators of state-of-the-art approaches are evaluated for comparison and the paper is concluded by discussing the trade-offs and notable remarks.

show abstract

Section: Research Challengesmentioning

confidence: 99%

Performance Analysis of Machine Learning Centered Workload Prediction Models for Cloud

Saxena,

Kumar,

Singh

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The clustering procedure groups slices into clusters such that slices that are 3 4 -monochromatic for the same color are all within the same cluster. In addition it must ensure that any cluster is not too large because the scheduling procedure will not split clusters among different servers and, thus, will not be able to find a good schedule if some clusters are very large.…”

Section: The Clustering Proceduresmentioning

confidence: 99%

“…Observation 4.11. Let S and T be 3 4 -monochromatic segments with majority color c. Then, S and T are both contained in the color c cluster. Lemma 4.12.…”

Section: Correctnessmentioning

confidence: 99%

“…• Monochromatic cost cost mono : when a slice S becomes 3 4 -monochromatic for some color c (and it was not 3 4 -monochromatic before) we move it to the color c cluster at a cost of |S|.…”

Section: Cost Of the Algorithmmentioning

confidence: 99%

“…Data-centric applications, including distributed machine learning, batch processing, scaleout databases, or streaming, produce a significant amount of communication traffic and their performance often critically depends on the underlying datacenter network [1]. In particular, large flows (also known as elephant flows) may require significant network resources if communicated across multiple hops, resulting in a high "bandwidth tax" and consuming valuable resources which would otherwise be available for additional flows [2,3].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Approximate Dynamic Balanced Graph Partitioning

Räcke

Schmid

Zabrodin

2022

Proceedings of the 34th ACM Symposium on Parallelism in Algorithms and Architectures

View full text Add to dashboard Cite

The performance of many large-scale and data-intensive distributed systems critically depends on the capacity of the interconnecting network. This paper is motivated by the vision of self-adjusting infrastructures whose resources can be adjusted according to the workload they currently serve, in a demand-aware manner. Such dynamic adjustments can be exploited to improve network utilization and hence performance, by dynamically moving frequently interacting communication partners closer, e.g., collocating them in the same server or datacenter rack.In particular, we revisit the online balanced graph partitioning problem which captures the fundamental tradeoff between the benefits and costs of dynamically collocating communication partners. The demand is modelled as a sequence σ (revealed in an online manner) of communication requests between n processes, each of which is running on one of the ℓ servers. Each server has capacity k = n/ℓ, hence, the processes have to be scheduled in a balanced manner across the servers. A request incurs cost 1, if the requested processes are located on different servers, otherwise the cost is 0. A process can be migrated to a different server at cost 1.This paper presents the first online algorithm for online balanced graph partitioning achieving a polylogarithmic competitive ratio for the fundamental case of ring communication patterns. Specifically, our main contribution is a O(log 3 n)-competitive randomized online algorithm for this problem. We further present a randomized online algorithm which is O(log 2 n)-competitive when compared to a static optimal solution. Our two results rely on different algorithms and techniques and hence are of independent interest.

show abstract

FMap: A fuzzy map for scheduling elephant flows throughjumping traveling salesman problemvariant towardsoftware‐defined networking‐baseddata center networks

Abdollahi

Asadi

Montazerolghaem

et al. 2023

Concurrency and Computation

View full text Add to dashboard Cite

SummaryNowadays data center networks (DCNs) should handle the ever‐growing load generated by diverse applications. It particularly occurs under concurrent flow requests and so‐called mice flows (MFs): elephant flows (EFs) ratio. Concentrating on a binary vision over flow size classification (EF or MF) results in subsequent unpredicted load imbalance due to neglecting EF's distinctions including a wide range of sizes. As a result, some EFs might utilize a path owing qualities beyond the given EF's demands, while another EF with higher requirements is attending to use an over‐utilized path. This article proposes FMap, a fuzzy map for scheduling EFs through our proposed variant of traveling salesperson problem (TSP) toward DCNs. FMap represents a novel EF scheduling scheme that integrates flow prioritization and routing decisions under the event pf parallel incoming flows besides the cooperation of the controller and OpenFlow switches in software‐defined networking (SDN) paradigm. FMap adopts fuzzy inference process to overcome the vagueness over EF's resource allocation. Mainly, FMap proposes a new variant of TSP (optimized by genetic algorithm) which enables EF's group forwarding with a minimum cost. FMap reduces the total hop count of EFs through considering a single optimal path for delivering groups of EFs that contain a same tag (priority). The outstanding results represent a major improvement as compared with equal cost multiple path, Hedera, Sonoum, and Size‐KP‐PSO. Particularly, the results illustrate an outperforming by 3.76×, 0.21×, 0.15×, 0.03×, and 0.03× in terms of total hop count, EFs FCT, packet loss, goodput, and received packets as compared with Size‐KP‐PSO, respectively.

show abstract

Cerberus

Cited by 28 publications

References 45 publications

Performance Analysis of Machine Learning Centered Workload Prediction Models for Cloud

Performance Analysis of Machine Learning Centered Workload Prediction Models for Cloud

Approximate Dynamic Balanced Graph Partitioning

FMap: A fuzzy map for scheduling elephant flows throughjumping traveling salesman problemvariant towardsoftware‐defined networking‐baseddata center networks

Contact Info

Product

Resources

About