scispace - formally typeset
Search or ask a question
Author

Amir Roozbeh

Other affiliations: Ericsson
Bio: Amir Roozbeh is an academic researcher from Royal Institute of Technology. The author has contributed to research in topics: Cloud computing & Page. The author has an hindex of 6, co-authored 18 publications receiving 111 citations. Previous affiliations of Amir Roozbeh include Ericsson.

Papers
More filters
Proceedings ArticleDOI
25 Mar 2019
TL;DR: A slice-aware memory management scheme, wherein frequently used data can be accessed faster via the LLC, and it is shown that a key-value store can potentially improve its average performance by up to 12.2% and 11.4% for 100% & 95% GET workloads, respectively.
Abstract: In modern (Intel) processors, Last Level Cache (LLC) is divided into multiple slices and an undocumented hashing algorithm (aka Complex Addressing) maps different parts of memory address space among these slices to increase the effective memory bandwidth. After a careful study of Intel's Complex Addressing, we introduce a slice-aware memory management scheme, wherein frequently used data can be accessed faster via the LLC. Using our proposed scheme, we show that a key-value store can potentially improve its average performance ~12.2% and ~11.4% for 100% & 95% GET workloads, respectively. Furthermore, we propose CacheDirector, a network I/O solution which extends Direct Data I/O (DDIO) and places the packet's header in the slice of the LLC that is closest to the relevant processing core. We implemented CacheDirector as an extension to DPDK and evaluated our proposed solution for latency-critical applications in Network Function Virtualization (NFV) systems. Evaluation results show that CacheDirector makes packet processing faster by reducing tail latencies (90-99th percentiles) by up to 119 μs (~21.5%) for optimized NFV service chains that are running at 100 Gbps. Finally, we analyze the effectiveness of slice-aware memory management to realize cache isolation.

56 citations

Proceedings Article
01 Jan 2020
TL;DR: Memory access is the major bottleneck in realizing multi-hundred-gigabit networks with commodity hardware, hence it is essential to make good use of cache memory that is a faster, but smaller memor ...
Abstract: Memory access is the major bottleneck in realizing multi-hundred-gigabit networks with commodity hardware, hence it is essential to make good use of cache memory that is a faster, but smaller memor ...

41 citations

Journal ArticleDOI
TL;DR: An overview of the functional architecture of a cloud built on SDHI is provided, exploring how the impact of this transformation goes far beyond the cloud infrastructure level in its impact on platforms, execution environments, and applications.
Abstract: This paper provides an overview of software-defined “hardware” infrastructures (SDHI). SDHI builds upon the concept of hardware (HW) resource disaggregation. HW resource disaggregation breaks today’s physical server-oriented model where the use of a physical resource (e.g., processor or memory) is constrained to a physical server’s chassis. SDHI extends the definition of of software-defined infrastructures (SDI) and brings greater modularity, flexibility, and extensibility to cloud infrastructures, thus allowing cloud operators to employ resources more efficiently and allowing applications not to be bounded by the physical infrastructure’s layout. This paper aims to be an initial introduction to SDHI and its associated technological advancements. This paper starts with an overview of the cloud domain and puts into perspective some of the most prominent efforts in the area. Then, it presents a set of differentiating use-cases that SDHI enables. Next, we state the fundamentals behind SDI and SDHI, and elaborate why SDHI is of great interest today. Moreover, it provides an overview of the functional architecture of a cloud built on SDHI, exploring how the impact of this transformation goes far beyond the cloud infrastructure level in its impact on platforms, execution environments, and applications. Finally, an in-depth assessment is made of the technologies behind SDHI, the impact of these technologies, and the associated challenges and potential future directions of SDHI.

34 citations

Proceedings ArticleDOI
19 Apr 2021
TL;DR: PacketMill as discussed by the authors is a system for optimizing software packet processing, which introduces a new model to efficiently manage packet metadata and employs code-optimization techniques to better utilize commodity hardware.
Abstract: We present PacketMill, a system for optimizing software packet processing, which (i) introduces a new model to efficiently manage packet metadata and (ii) employs code-optimization techniques to better utilize commodity hardware. PacketMill grinds the whole packet processing stack, from the high-level network function configuration file to the low-level userspace network (specifically DPDK) drivers, to mitigate inefficiencies and produce a customized binary for a given network function. Our evaluation results show that PacketMill increases throughput (up to 36.4 Gbps -- 70%) & reduces latency (up to 101 us -- 28%) and enables nontrivial packet processing (e.g., router) at ~100 Gbps, when new packets arrive >10× faster than main memory access times, while using only one processing core.

24 citations

Proceedings ArticleDOI
07 Dec 2015
TL;DR: It is beneficial to move session management to data centers collocated with the BS on 5G network when there is high user density, and an analysis of the control signaling performance for each proposed control function is presented.
Abstract: 5G is the next generation of mobile network. There are many requirements on 5G, such as high capacity, low latency, flexibility, and support for any-to-any communication. Cloud technology, in the form of a distributed cloud (also known as a network embedded cloud), is an enabler technology for 5G by allowing flexible networks that meet different user application requirements. On the other hand, Machine Type Communication (MTC) is a primary application for 5G, but it can add a high volume of control signaling. To manage the expected high volume of control signaling introduced by MTC, we identified the main control events that generate signaling messages in the network. Then, we proposed a decentralized core network architecture optimized for the identified control events. The proposed control plane functions are independent in the sense that each can be executed separately. The control functions can utilize the distributed cloud to manage the enormous amount of control signaling by handling this signaling locally. Additionally, we present an analysis of the control signaling performance for each proposed control function. We conclude that it is beneficial to move session management to data centers collocated with the BS on 5G network when there is high user density.

8 citations


Cited by
More filters
Journal ArticleDOI
TL;DR: This survey aims at providing a comprehensive survey of state-of-the-art research work, which leverages SDN and NFV into the most recent mobile packet core network architecture, evolved packet core.
Abstract: The emergence of two new technologies, namely, software defined networking (SDN) and network function virtualization (NFV), have radically changed the development of network functions and the evolution of network architectures. These two technologies bring to mobile operators the promises of reducing costs, enhancing network flexibility and scalability, and shortening the time-to-market of new applications and services. With the advent of SDN and NFV and their offered benefits, the mobile operators are gradually changing the way how they architect their mobile networks to cope with ever-increasing growth of data traffic, massive number of new devices and network accesses, and to pave the way toward the upcoming fifth generation networking. This survey aims at providing a comprehensive survey of state-of-the-art research work, which leverages SDN and NFV into the most recent mobile packet core network architecture, evolved packet core. The research work is categorized into smaller groups according to a proposed four-dimensional taxonomy reflecting the: 1) architectural approach, 2) technology adoption, 3) functional implementation, and 4) deployment strategy. Thereafter, the research work is exhaustively compared based on the proposed taxonomy and some added attributes and criteria. Finally, this survey identifies and discusses some major challenges and open issues, such as scalability and reliability, optimal resource scheduling and allocation, management and orchestration, and network sharing and slicing that raise from the taxonomy and comparison tables that need to be further investigated and explored.

269 citations

Journal ArticleDOI
TL;DR: The most recent standardization activities on IAB are described, and architectures with and without IAB in mmWave deployments are compared, to demonstrate the cell edge throughput advantage offered by IAB using endto- end system-level simulations.
Abstract: IAB is being considered as a means to reduce the deployment costs of ultra-dense 5G mmWave networks, using wireless backhaul links to relay the access traffic. In this work we describe the most recent standardization activities on IAB, and compare architectures with and without IAB in mmWave deployments. While it is well understood that IAB networks reduce deployment costs by obviating the need to provide wired backhaul to each cellular base station, it is still necessary to validate the IAB performance in realistic scenarios. In this article we demonstrate the cell edge throughput advantage offered by IAB using endto- end system-level simulations. We also highlight some research challenges for IAB that will require further investigations.

119 citations

Journal ArticleDOI
TL;DR: This article investigates the problem of reliable resource provisioning in joint edge-cloud environments, and surveys technologies, mechanisms, and methods that can be used to improve the reliability of distributed applications in diverse and heterogeneous network environments.
Abstract: Large-scale software systems are currently designed as distributed entities and deployed in cloud data centers. To overcome the limitations inherent to this type of deployment, applications are increasingly being supplemented with components instantiated closer to the edges of networks—a paradigm known as edge computing. The problem of how to efficiently orchestrate combined edge-cloud applications is, however, incompletely understood, and a wide range of techniques for resource and application management are currently in use. This article investigates the problem of reliable resource provisioning in joint edge-cloud environments, and surveys technologies, mechanisms, and methods that can be used to improve the reliability of distributed applications in diverse and heterogeneous network environments. Due to the complexity of the problem, special emphasis is placed on solutions to the characterization, management, and control of complex distributed applications using machine learning approaches. The survey is structured around a decomposition of the reliable resource provisioning problem into three categories of techniques: workload characterization and prediction, component placement and system consolidation, and application elasticity and remediation. Survey results are presented along with a problem-oriented discussion of the state-of-the-art. A summary of identified challenges and an outline of future research directions are presented to conclude the article.

100 citations

Posted Content
TL;DR: In this article, the authors demonstrate the cell-edge throughput advantage offered by integrated access and backhaul (IAB) using end-to-end system level simulations, and highlight some research challenges associated with this architecture that will require further investigations.
Abstract: Integrated Access and Backhaul (IAB) is being investigated as a means to overcome deployment costs of ultra-dense 5G millimeter wave (mmWave) networks by realizing wireless backhaul links to relay the access traffic. For the development of these systems, however, it is fundamental to validate the performance of IAB in realistic scenarios through end-to-end system level simulations. In this paper, we shed light on the most recent standardization activities on IAB, and compare architectures with and without IAB in mmWave deployments. While it is well understood that IAB networks reduce deployment costs by obviating the need to provide wired backhaul to each cellular base-station, in this paper we demonstrate the cell-edge throughput advantage offered by IAB using end-to-end system level simulations. We further highlight some research challenges associated with this architecture that will require further investigations.

82 citations

Proceedings ArticleDOI
09 Aug 2021
TL;DR: In this article, the authors present measurement and insights for Linux kernel network stack performance for 100Gbps access link bandwidths and find that such high bandwidth links, coupled with relatively stagnant technology trends for other host resources (e.g., CPU speeds and capacity, cache sizes, NIC buffer sizes, etc.), mark a fundamental shift in host network stack bottlenecks.
Abstract: Traditional end-host network stacks are struggling to keep up with rapidly increasing datacenter access link bandwidths due to their unsustainable CPU overheads. Motivated by this, our community is exploring a multitude of solutions for future network stacks: from Linux kernel optimizations to partial hardware offload to clean-slate userspace stacks to specialized host network hardware. The design space explored by these solutions would benefit from a detailed understanding of CPU inefficiencies in existing network stacks. This paper presents measurement and insights for Linux kernel network stack performance for 100Gbps access link bandwidths. Our study reveals that such high bandwidth links, coupled with relatively stagnant technology trends for other host resources (e.g., CPU speeds and capacity, cache sizes, NIC buffer sizes, etc.), mark a fundamental shift in host network stack bottlenecks. For instance, we find that a single core is no longer able to process packets at line rate, with data copy from kernel to application buffers at the receiver becoming the core performance bottleneck. In addition, increase in bandwidth-delay products have outpaced the increase in cache sizes, resulting in inefficient DMA pipeline between the NIC and the CPU. Finally, we find that traditional loosely-coupled design of network stack and CPU schedulers in existing operating systems becomes a limiting factor in scaling network stack performance across cores. Based on insights from our study, we discuss implications to design of future operating systems, network protocols, and host hardware.

58 citations