Jericho: Achieving scalability through optimal data placement on multicore systems

Mavridis, Stelios; Sfakianakis, Yannis; Papagiannis, Anastasios; Marazakis, Manolis; Bilas, Angelos

doi:10.1109/msst.2014.6855538

Cited by 6 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indirect LLC Isolation: FRIMICS does not explicitly isolate CPU caches (specifically the LLC) and memory bandwidth; however, it mitigates the interference in those resources by placing applications to a single NUMA node, when there are enough resources available, or uses neighboring NUMA nodes when applications are larger. Therefore, it indirectly reduces interference in percore caches and directly traffic across different NUMA nodes, which can significantly improve I/O throughput up to 2× [30].…”

Section: Frimicsmentioning

confidence: 99%

QuMan

Sfakianakis

Kozanitis

Kozyrakis

et al. 2018

ACM Trans. Archit. Code Optim.

Self Cite

View full text Add to dashboard Cite

Modern data centers consolidate workloads to increase server utilization and reduce total cost of ownership, and cope with scaling limitations. However, server resource sharing introduces performance interference across applications and, consequently, increases performance volatility, which negatively affects user experience. Thus, a challenging problem is to increase server utilization while maintaining application QoS. In this article, we present QuMan, a server resource manager that uses application isolation and profiling to increase server utilization while controlling degradation of application QoS. Previous solutions, either estimate interference across applications and then restrict colocation to "compatible" applications, or assume that application requirements are known. Instead, QuMan estimates the required resources of applications. It uses an isolation mechanism to create properly-sized resource slices for applications, and arbitrarily colocates applications. QuMan's mechanisms can be used with a variety of admission control policies, and we explore the potential of two such policies: (1) A policy that allows users to specify a minimum performance threshold and (2) an automated policy, which operates without user input and is based on a new combined QoS-utilization metric. We implement QuMan on top of Linux servers, and we evaluate its effectiveness using containers and real applications. Our single-node results show that QuMan balances highly effectively the tradeoff between server utilization and application performance, as it achieves 80% server utilization while the performance of each application does not drop below 80% the respective standalone performance. We also deploy QuMan on a cluster of 100 AWS instances that are managed by a modified version of the Sparrow scheduler [37] and, we observe a 48% increase in application performance on a highly utilized cluster, compared to the performance of the same cluster under the same load when it is managed by native Sparrow or Apache Mesos.

show abstract

Section: Frimicsmentioning

confidence: 99%

QuMan

Sfakianakis

Kozanitis

Kozyrakis

et al. 2018

ACM Trans. Archit. Code Optim.

Self Cite

View full text Add to dashboard Cite

show abstract

“…In [16] using Vanguard, we show the improvement in memory throughput when applying task placement, where we minimize NUMA effects in the system.…”

Section: Journaling Filesystemmentioning

confidence: 99%

“…pCache [16] provides a partitioned DRAM I/O cache to reduce memory contention due to global policies and to allow NUMA placement. In typical servers today all workloads use the shared Linux page cache.…”

Section: Dram I/o Cachementioning

confidence: 99%

Vanguard

Sfakianakis

Mavridis

Papagiannis

et al. 2014

Proceedings of the ACM Symposium on Cloud Computing

Self Cite

View full text Add to dashboard Cite

Server consolidation via virtualization is an essential technique for improving infrastructure cost in modern datacenters. From the viewpoint of datacenter operators, consolidation offers compelling advantages by reducing the number of physical servers, and reducing operational costs such as energy consumption. However, performance interference between co-located workloads can be crippling. Conservatively, and at significant cost, datacenter operators are forced to keep physical servers at low utilization levels (typically below 20%), to minimize adverse performance interactions.In this paper, we focus on addressing the issue of performance interference on a virtualized server operating at high utilization levels. In our work, we find that eliminating interference in the I/O path is critical for achieving good performance on consolidated servers. We present Vanguard, a device driver stack that implements a full I/O path in the Linux kernel that provisions competing workloads with dedicated resources. We focus on two key resources: in-memory buffers for the filesystem, and space on SSD devices that serve as a transparent cache for block devices. Our approach effectively mitigates performance interference, for several mixes of transactional, streaming, and analytical processing workloads. We find that with our approach a server can run more workloads close to their nominal performance level as compared to the unmodified Linux I/O path, by careful allocation of I/O path resources to each workload. At excessive load levels, i.e. when the aggregate load exceeds the capa-0 † Also, with the bilities of the server, our approach can still provide isolated slices of the I/O path for a subset of the co-located workloads yielding at least 50% of their nominal performance. In addition, Vanguard is shown to be 2.5x more efficient in terms of service resource usage for a 4-workload mix, taking into account utilization and power consumption. With an I/O-heavy mix 6-workload mix, Vanguard is 8x more efficient than the unmodified baseline Linux system.

show abstract

Understanding the performance of storage class memory file systems in the NUMA architecture

Kim

Khan

et al. 2018

Cluster Comput

View full text Add to dashboard Cite

Jericho: Achieving scalability through optimal data placement on multicore systems

Cited by 6 publications

References 7 publications

QuMan

QuMan

Vanguard

Understanding the performance of storage class memory file systems in the NUMA architecture

Contact Info

Product

Resources

About