Corrigendum to “The IX Operating System: Combining Low Latency, High Throughput and Efficiency in a Protected Dataplane”

Belay, Adam; Prekas, George; Primorac, Mia; Klimovic, Ana; Grossman, Samuel; Kozyrakis, Christos; Bugnion, Edouard

doi:10.1145/3154292

Cited by 24 publications

(46 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…is focus is done at the expense of a description of some more advanced research concepts. For example, the text does not discuss recursive virtual machines [33,158], the use of virtualization hardware for purposes other than running traditional virtual machines [24,29,31,43,88], or the emerging question of architectural support for containers such as Docker [129].…”

Section: Organization Of This Bookmentioning

confidence: 99%

“…In 2005, he co-founded Nuova Systems, a hardware company premised on providing architectural support for virtualization in the network and the I/O subsystem, which became the core of Cisco's Data Center strategy. More recently, having returned to academia as a professor at École polytechnique fédérale de Lausanne (EPFL), Edouard is now involved in the IX project [30,31,147] which leverages virtualization hardware and the Dune framework [29] to build specialized operating systems.…”

Section: Authors' Perspectivesmentioning

confidence: 99%

See 1 more Smart Citation

Hardware and Software Support for Virtualization

Bugnion

Nieh

Tsafrir

2017

Synthesis Lectures on Computer Architecture

Self Cite

View full text Add to dashboard Cite

show abstract

Section: Organization Of This Bookmentioning

confidence: 99%

Section: Authors' Perspectivesmentioning

confidence: 99%

Hardware and Software Support for Virtualization

Bugnion

Nieh

Tsafrir

2017

Synthesis Lectures on Computer Architecture

Self Cite

View full text Add to dashboard Cite

show abstract

“…The former can efficiently schedule the resources of a multi-core server and prioritize latency-sensitive tasks [8] but suffers from high overheads for µs-scale tasks. The latter improves throughput substantially (by up to 6× for key-value stores [5]) through sweeping simplifications such as separation of control from the dataplane execution, polling, run-to-completion, and synchronization-free, flow-consistent mapping of requests to cores [5,26,27,39,42,51].…”

Section: Introductionmentioning

confidence: 99%

“…These sweeping simplifications lead to two related forms of inefficiencies: (a) the dataplane is not a work conserving scheduler, i.e., a core may be idle while there are pending requests; and (b) the dataplane suffers from head-of-line blocking, i.e., a request may be blocked until the previous tasks complete execution. While these limitations might be acceptable to workloads with near-deterministic task execution time and relatively loose SLO (e.g., some widely-studied memcached workloads [1,43] with an SLO at > 100× the mean service time [5]), such assumptions break down when considering more complex workloads, e.g., in-memory transaction processing with a TPC-C-like mix of requests or with more aggressive SLO targets.…”

Section: Introductionmentioning

confidence: 99%

“…(4) We compare ZYGOS to IX, a state-of-the-art dataplane with strict run-to-completion that partitions flows onto cores [5]. While ZYGOS's scheduler introduces some necessary buffering, communication and synchronization (which are measurable for extremely small tasks), it eliminates headof-line blocking and clearly outperforms IX for tasks ≥10µs ( §6.1).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ZygOS

Prekas

Kogias

Bugnion

2017

Proceedings of the 26th Symposium on Operating Systems Principles

Self Cite

135

View full text Add to dashboard Cite

This paper focuses on the efficient scheduling on multicore systems of very fine-grain networked tasks, which are the typical building block of online data-intensive applications. The explicit goal is to deliver high throughput (millions of remote procedure calls per second) for tail latency servicelevel objectives that are a small multiple of the task size. We present ZYGOS, a system optimized for µs-scale, inmemory computing on multicore servers. It implements a work-conserving scheduler within a specialized operating system designed for high request rates and a large number of network connections. ZYGOS uses a combination of shared-memory data structures, multi-queue NICs, and inter-processor interrupts to rebalance work across cores. For an aggressive service-level objective expressed at the 99 t h percentile, ZYGOS achieves 75% of the maximum possible load determined by a theoretical, zero-overhead model (centralized queueing with FCFS) for 10µs tasks, and 88% for 25µs tasks. We evaluate ZYGOS with a networked version of Silo, a state-of-the-art in-memory transactional database, running TPC-C. For a service-level objective of 1000µs latency at the 99 t h percentile, ZYGOS can deliver a 1.63× speedup over Linux (because of its dataplane architecture) and a 1.26× speedup over IX, a state-of-the-art dataplane (because of its work-conserving scheduler).

show abstract

5G‐oriented non‐orthogonal multiple access technology

Zhang

Yang

2020

Int J Communication

View full text Add to dashboard Cite

Summary With the changes of the times and the rapid development of the mobile Internet and the Internet of Things, network communications are facing increasingly severe challenges. Due to the user exclusivity of the orthogonal resources of the existing orthogonal multiple access technology, it is difficult for the number of accessible users and the spectral efficiency to meet the application requirements of network communications. 5G technology will meet the diverse business needs of people in various scenarios such as life, work, leisure, and transportation. This paper analyzes the 5G‐oriented non‐orthogonal multiple access (NOMA) technology and designs and implements a NOMA technology simulation platform laboratory based on 5G technology. The entire simulation experiment adopts a modular design concept so that it can be modified and extended in the future network communication. In this paper, the user pairing scheme and power allocation algorithm of NOMA technology are studied in depth. For the user pairing scheme, this paper adopts two schemes: traversal search pairing scheme and grouping pairing scheme. For the same algorithm for power allocation, this paper studies two suboptimal power allocation algorithms including fixed power allocation and partial transmission power allocation. Simulation experiments show that the average cell throughput and cell edge user throughput of NOMA are slightly lower than when the traversal search pairing scheme is adopted: the average cell throughput has decreased by approximately 4%, and the cell edge user throughput has decreased by approximately 1.6%. The packet pairing scheme is a compromise between complexity and throughput performance. For different power allocation factor values, the overall throughput performance of some transmission power allocation algorithms is better than the flower pollination algorithm (FPA).

show abstract

Corrigendum to “The IX Operating System: Combining Low Latency, High Throughput and Efficiency in a Protected Dataplane”

Cited by 24 publications

References 0 publications

Hardware and Software Support for Virtualization

Hardware and Software Support for Virtualization

ZygOS

5G‐oriented non‐orthogonal multiple access technology

Contact Info

Product

Resources

About