Bandwidth-Aware Page Placement in NUMA

Gureya, David; Neto, João Paulo; Karimi, Reza; Barreto, João; Bhatotia, Pramod; Quéma, Vivien; Rodrigues, Rodrigo; Romano, Paolo; Vlassov, Vladimir

doi:10.1109/ipdps47924.2020.00063

Cited by 11 publications

(10 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When E = ∅, COLOURED-BIN-PACKING reduces to the standard bin packing problem (d = 1). Therefore, the problem is NP-complete, APX-hard and no algorithm can achieve a better approximation ratio than 3 2 (unless P=NP). It comes from the approximation hardness of standard bin packing through a reduction from the partition problem [18].…”

Section: E Colored Bin Packingmentioning

confidence: 99%

“…In particular, the network infrastructure across Internet Service Providers consists of compute servers made up of different capacity constraints zones as part of the horizontal scaling. An example of such zones are the NUMA zones [3] which are used as a protection mechanism for the host and are practically being deployed as such on physical motherboards.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A multidimensional colored packing approach for network slicing with dedicated protection

Lesfari

Giroire

Lena

et al. 2021

2021 IEEE Global Communications Conference (GLOBECOM)

View full text Add to dashboard Cite

Network Function Virtualization (NFV) enables the virtualization of core-business network functions on top of a NFV infrastructure. NFV has gained an increasing attention in the telecommunication field these last few years. Virtual network functions (VNFs) can be represented by a set of virtual network function components (VNFCs). These VNFCs are typically designed with a redundancy scheme and need to be deployed against failures of, e.g., compute servers. However, such deployment must respect a particular resiliency mechanism for protection purposes. Therefore, choosing an efficient mapping of VNFCs to the compute servers is a challenging problem in the optimization of the software-defined, virtualization-based next generation of networks. In this paper, we model the problem of reliable VNFCs placement under anti-affinity constraints using several optimization techniques. A novel approach based on an extension of bin packing is proposed. We perform a comprehensive evaluation in terms of performance under realworld ISP networks along with synthetic traces. We show that our methods can calculate rapidly efficient solutions for large instances.

show abstract

Section: E Colored Bin Packingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A multidimensional colored packing approach for network slicing with dedicated protection

Lesfari

Giroire

Lena

et al. 2021

2021 IEEE Global Communications Conference (GLOBECOM)

View full text Add to dashboard Cite

show abstract

“…Therefore, to quantify the potential advantage of an ideal bandwidth balance, we return to our initial benchmark, and focus exclusively on the all reads workload -i.e., the best-case scenario for bandwidth balance. In contrast to the previous experiment, this time we distribute pages across tiers according to different ratios (100% in DRAM, 95%, 90%, ...), using weighted-interleaved placement [15]. Also, this time we vary the number of active threads to test different memory access rates.…”

Section: Bandwidth Balance Policymentioning

confidence: 99%

Dynamic Page Placement on Real Persistent Memory Systems

Marques¹,

Kuzmin²,

Barreto³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

As persistent memory (PM) technologies emerge, hybrid memory architectures combining DRAM with PM bring the potential to provide a tiered, byte-addressable main memory of unprecedented capacity. Nearly a decade after the first proposals for these hybrid architectures, the real technology has finally reached commercial availability with Intel Optane™ DC Persistent Memory (DCPMM). This raises the challenge of designing systems that realize this potential in practice, namely through effective approaches that dynamically decide at which memory tier should pages be placed.In this paper, we are the first, to our knowledge, to systematically analyze tiered page placement on real DCPMM-based systems. To this end, we start by revisiting the assumptions of state-of-the-art proposals, and confronting them with the idiosyncrasies of today's off-the-shelf DCPMM-equipped architectures. This empirical study reveals that some of the key design choices in the literature rely on important assumptions that are not verified in present-day DRAM-DCPMM memory architectures.Based on the lessons from this study, we design and implement HyPlacer, a tool for tiered page placement in off-the-shelf Linux-based systems equipped with DRAM+DCPMM. In contrast to previous proposals, HyPlacer follows an approach guided by two main practicality principles: 1) it is tailored to the performance idiosyncrasies of off-theshelf DRAM+DCPMM systems; and 2) it can be seamlessly integrated into Linux with minimal kernel-mode components, while ensuring extensibility to other HMAs and other data placement policies. Our experimental evaluation of HyPlacer shows that it outperforms both solutions proposed in past literature and placement options that are currently available in off-the-shelf DCPMM-equipped Linux systems, reaching an improvement of up to 11x when compared to the default memory policy in Linux.

show abstract

“…State of the art memory pages placement mechanisms interleave pages evenly across NUMA nodes. However, this approach fails to maximize memory throughput in NUMA systems characterized by asymmetric bandwidths and latencies, and sensitive to memory contention and interconnection congestion effects 9 . Different solutions have been proposed, like modifying the Linux kernel for improving load balancing and thread migration algorithms, changing memory pages placement policies, or profiling the executed program for obtaining the optimal thread placement, among others.…”

Section: Related Workmentioning

confidence: 99%

“…They have obtained important improvements in performance with PARSEC 3.0 benchmarks. Also, the most recent work by Gureya et al 9 proposes a novel page placement mechanism based on asymmetric weight page interleaving that combines an analytical performance model of the target NUMA systems with on‐line iterative tuning page distribution for a given memory‐intensive application. By only migrating memory pages and entrusting thread migration to the OS, they obtain up to a 66% performance improvement.…”

Section: Related Workmentioning

confidence: 99%

LBMA and IMAR²: Weighted lottery based migration strategies for NUMA multiprocessing servers

Laso

Lorenzo

Rivera

et al. 2020

Concurrency and Computation

View full text Add to dashboard Cite

Summary Multicore NUMA systems present on‐board memory hierarchies and communication networks that influence performance when executing shared memory parallel codes. Characterizing this influence is complex, and understanding the effect of particular hardware configurations on different codes is of paramount importance. In this article, monitoring information extracted from hardware counters at runtime is used to characterize the behavior of each thread for an arbitrary number of multithreaded processes running in a multiprocessing environment. This characterization is given in terms of number of operations per second, operational intensity, and latency of memory accesses. We propose a runtime tool, executed in user space, that uses this information to guide two different thread migration strategies for improving execution efficiency by increasing locality and affinity without requiring any modification in the running codes. Different configurations of NAS Parallel OpenMP benchmarks running concurrently on multicore NUMA systems were used to validate the benefits of our proposal, in which up to four processes are running simultaneously. In more than the 95% of the executions of our tool, results outperform those of the operating system (OS) and produces up to 38% improvement in execution time over the OS for heterogeneous workloads, under different and realistic locality and affinity scenarios.

show abstract

Bandwidth-Aware Page Placement in NUMA

Cited by 11 publications

References 29 publications

A multidimensional colored packing approach for network slicing with dedicated protection

A multidimensional colored packing approach for network slicing with dedicated protection

Dynamic Page Placement on Real Persistent Memory Systems

LBMA and IMAR²: Weighted lottery based migration strategies for NUMA multiprocessing servers

Contact Info

Product

Resources

About

Bandwidth-Aware Page Placement in NUMA

Cited by 11 publications

References 29 publications

A multidimensional colored packing approach for network slicing with dedicated protection

A multidimensional colored packing approach for network slicing with dedicated protection

Dynamic Page Placement on Real Persistent Memory Systems

LBMA and IMAR2: Weighted lottery based migration strategies for NUMA multiprocessing servers

Contact Info

Product

Resources

About

LBMA and IMAR²: Weighted lottery based migration strategies for NUMA multiprocessing servers