On the Fly TCP Acceleration with Miniproxy

Siracusano, Giuseppe; Bifulco, Roberto; Kuenzer, Simon; Melazzi, Nicola Blefari; Huici, Felipe

doi:10.1145/2940147.2940149

Cited by 17 publications

(16 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our evaluation compares against netmap directly since our testing suggests it generally has higher performance. Their work, and others such as Jitsu [31] and Miniproxy [32], illustrate the potential of unikernels to reduce NF startup time to the 20 ms range. Flurries relies on an idle pool of NFs and we have not yet explored ways to reduce the startup delay of our NFs (approximately 500ms).…”

Section: Related Workmentioning

confidence: 98%

Flurries

Zhang

Hwang

Rajagopalan

et al. 2016

Proceedings of the 12th International on Conference on Emerging Networking EXperiments and Technologies

View full text Add to dashboard Cite

The combination of Network Function Virtualization (NFV) and Software Defined Networking (SDN) allows flows to be flexibly steered through efficient processing pipelines. As deployment of NFV becomes more prevalent, the need to provide fine-grained customization of service chains and flowlevel performance guarantees will increase, even as the diversity of Network Functions (NFs) rises. Existing NFV approaches typically route wide classes of traffic through preconfigured service chains. While this aggregation improves efficiency, it prevents flexibly steering and managing performance of flows at a fine granularity. To provide both efficiency and flexibility, we present Flurries, an NFV platform designed to support large numbers of short-lived lightweight NFs, potentially running a unique NF for each flow. Flurries maintains a pool of Docker container NFs-several thousand on each host-and resets NF memory state between flows for fast reuse. Flurries uses a hybrid of polling and interrupts to improve throughput and latency while allowing multiple NFs to efficiently share CPU cores. By assigning each NF an individual flow or a small set of flows, it becomes possible to dynamically manage the QoS and service chain functionality for flows at a very fine granularity. Our Flurries prototype demonstrates the potential for this approach to run as many as 80,000 Flurry NFs during a one second interval, while forwarding over 30Gbps of traffic, dramatically increasing data plane customizability.

show abstract

Section: Related Workmentioning

confidence: 98%

Flurries

Zhang

Hwang

Rajagopalan

et al. 2016

Proceedings of the 12th International on Conference on Emerging Networking EXperiments and Technologies

View full text Add to dashboard Cite

show abstract

“…However, as major disadvantages compared to our proposal this leads to a significant load on the proxy server and breaks the principle of end-to-end transport encryption between the client and the web server. Furthermore, Miniproxy [24] can be used to accelerate TCP's connection establishment. This approach places a proxy between the client and the web server, which doubles the number of required TCP handshakes.…”

Section: Related Workmentioning

confidence: 99%

Accelerating QUIC’s Connection Establishment on High-Latency Access Networks

Mueller

Moennich

et al. 2019

2019 IEEE 38th International Performance Computing and Communications Conference (IPCCC)

View full text Add to dashboard Cite

A significant amount of connection establishments on the web require a prior domain name resolution by the client. Especially on high-latency access networks, these DNS lookups cause a significant delay on the client's connection establishment with a server. To reduce the overhead of QUIC's connection establishment with prior DNS lookup on these networks, we propose a novel QuicSocks proxy. Basically, the client delegates the domain name resolution towards the QuicSocks proxy. Our results indicate, that colocating our proxy with real-world ISPprovided DNS resolvers provides great performance gains. For example, 10% of our 474 sample nodes distributed across ISP's in Germany would save at least 30 ms per QUIC connection establishment. The design of our proposal aims to be readily deployable on the Internet by avoiding IP address spoofing, anticipating Network Address Translators and using the standard DNS and QUIC protocols. In summary, our proposal fosters a faster establishment of QUIC connections for clients on highlatency access networks.

show abstract

“…For a given proxy path, the TCP transfer time during Slow-Start is modeled as a linear combination of the path length (i.e., sum of RTTs) and the maximum RTT within the path, i.e., between any pair of proxies [44]. The coefficients of this linear combination depend on the size of the data transfer, hence, different transfer sizes can have different optimal paths.…”

Section: B Computing Proxy-to-proxy Chainsmentioning

confidence: 99%

“…We build MOSTO proxies using Miniproxy [44], a Xen [5] unikernel that implements a high-performance, lightweight TCP proxy. The details of Miniproxy's implementation are presented in [44], here we just report that it boots in 12ms and requires as little as 6MBs of RAM to run. For the implementation of MOSTO, we added support for TCP segmentation offloading to the Miniproxy's network stack, achieving line rate 10Gbps throughput on a single core when proxying TCP connections 6 .…”

Section: B Tcp Proxy and Offloadingmentioning

confidence: 99%

See 1 more Smart Citation

Re-Designing Dynamic Content Delivery in the Light of a Virtualized Infrastructure

Siracusano

Bifulco²,

Trevisan

et al. 2017

IEEE J. Select. Areas Commun.

Self Cite

View full text Add to dashboard Cite

We explore the opportunities and design options enabled by novel SDN and NFV technologies, by re-designing a dynamic Content Delivery Network (CDN) service. Our system, named MOSTO, provides performance levels comparable to that of a regular CDN, but does not require the deployment of a large distributed infrastructure. In the process of designing the system, we identify relevant functions that could be integrated in the future Internet infrastructure. Such functions greatly simplify the design and effectiveness of services such as MOSTO. We demonstrate our system using a mixture of simulation, emulation, testbed experiments and by realizing a proof-of-concept deployment in a planet-wide commercial cloud system.Index Terms-NFV, Cloud, Edge Cloud, SDN, CDN, TCP proxy.performance of dynamic web content delivery from an origin server. The motivation for selecting a CDN as application is twofold. First, CDNs are widely deployed and play a critical role in today's Internet [37]. Second, building a CDN requires a fine engineering work to balance performance, scalability and costs. Complex algorithms for content placement and routing [32], fast software implementations [27], tuning of network protocols [3] are just few of the required building blocks, which allow us to touch a large variety of topics. Furthermore, the decision to target dynamic content delivery focuses our discussion on networking-related issues, rather than on cache dynamics as it is the case for static content delivery [49].MOSTO dynamically places virtualized proxies in the network to accelerate TCP flows. In effect, we revisit a wellknown technique for TCP acceleration [30], i.e., using a chain of TCP proxies to connect the end-points, in the light of recent advances on network function virtualization [35] and programmable network devices [12]. In particular, we provide two main contributions. First, we describe the way MOSTO addresses the following issues:• Mapping clients to ingress proxies and intercepting client network traffic; • Selecting proxies' locations and proxies chains for a given TCP flow; • Scaling proxies forwarding throughput to minimize the cost of running the system.

show abstract

On the Fly TCP Acceleration with Miniproxy

Cited by 17 publications

References 19 publications

Flurries

Flurries

Accelerating QUIC’s Connection Establishment on High-Latency Access Networks

Re-Designing Dynamic Content Delivery in the Light of a Virtualized Infrastructure

Contact Info

Product

Resources

About