Most of the work on VM network performance has focused so far bulk TCP traffic, which covers classical applications virtualization. Completely new “paravirtualized devices” (Xenfront, VIRTIO, vmxnet) have been designed and implemented to improve throughput. We expect virtualization become widely used also for different workloads: packet switching devices middleboxes, Software Defined Networks, etc.. These involve very high rates that are problematic not only hypervisor (which emulates...

The rising interest in Network Function Virtualization (NFV) requires Virtual Machines (VMs) to operate with diversified networking workloads, from traditional, bulk TCP transfers novel ones featuring extremely high packet rates. In response, researchers have explored and proposed new solutions for performance VM networking, including optimizations virtual network adapters (such as VirtIO) support speed traffic, alternative frameworks userspace physical or passthrough. To date, we are still...

The Recursive InterNetwork Architecture (RINA) is a general architecture for all forms of computer networking, based on single type programmable layer that recurs as many times required by the network designer. recursion and programmability aspects RINA are key to design flexible, heterogeneous networks while still bounding their complexity. In this paper we show how enabled can be exploited in practice means Software Development Kit (SDK) developed IRATI, open source implementation. A proof...

Most of the work on VM network performance has focused so far bulk TCP traffic, which covers classical applications virtualization. Completely new "paravirtualized devices" (Xenfront, VIRTIO, vmxnet) have been designed and implemented to improve throughput. We expect virtualization become widely used also for different workloads: packet switching devices middleboxes, Software Defined Networks, etc.. These involve very high rates that are problematic not only hypervisor (which emulates...

Network Function Virtualization has been touted as the silver bullet for tackling a number of operator problems, including vendor lock-in, fast deployment new functionality, converged management, and lower expenditure since packet processing runs on inexpensive commodity servers. The reality, however, is that, in practice, it proved hard to achieve stable, predictable performance provided by hardware middleboxes, so operators have essentially resorted throwing money at problem, deploying...

Tenants in a cloud environment run services, such as Virtual Network Function instantiations, that may legitimately generate millions of packets per second. The hosting platform, hence, needs robust packet scheduling mechanisms support these rates and, at the same time, provide isolation and dependable service guarantees under all load conditions. Current hardware or software solutions fail to meet requirements, most commonly lacking on either performance guarantees. In this paper we propose...

In this study, we investigate some counterintuitive but frequent performance issues that arise when doing high-speed networking (or I/O in general) with Virtual Machines (VMs). VMs use one or more single-producer/single-consumer systems to exchange data (e.g. network packets) their hypervisor. We show the producer and consumer process packets at different rates, high cost required for synchronization (interrupts 'kicks') may reduce throughput of system well below slowest two parties;...

This work addresses an apparently simple but elusive problem that arises when doing high speed networking on Virtual Machines. When a VM and its peer (usually the hypervisor) process packets at different rates, required for synchronization (interrupts "kicks") may reduce throughput well below slowest of two parties.

The Recursive InterNetwork Architecture (RINA) is a recently proposed network architecture based on first principles, which promises to solve number of issues present in the current Internet such as lack inherent security. In this paper, we experimental evaluation performance-oriented implementation RINA, IRATI stack. Our open source stack designed for GNU/Linux Operating Systems, with key components developed kernel space optimal performance. After briefly introducing architecture, main...

Virtualization is an enabling technology that improves scalability, reliability, and flexibility. Virtualized networking tackled by emulating or paravirtualizing network interface cards. This approach, however, leads to complexities (implementation management) has conform some limitations imposed the Ethernet standard. RINA turns current approach virtualized on its head: instead of networks perform inter-process communication a single processing system, it sees as extension local...

Summary Data processing pipelines normally use lockless Single‐Producer–Single‐Consumer (SPSC) queues to efficiently decouple their threads and achieve high throughput, minimizing the cost of synchronization. SPSC have been widely studied, mostly for applications such as streaming data or network monitoring, where main goal is maximizing throughput. There are now many applications, virtual‐machine–virtual‐machine communication, software‐defined networking, message‐based kernels, low latency...

A number of recent EU-funded projects have been investigating the Recursive Internet Architecture (RINA). IRATI built an initial prototype implementation, which was extended by PRISTINE project towards technology demonstrators showing feasibility architecture and demonstrating how RINA tackles security reliability it can simplify network management. Currently, ARCFIRE sets out to evaluate realistic scenarios, scaling up experiments in terms numbers nodes, services running time. In this paper...

IRATI is a open source implementation of RINA for OS/Linux systems that allows researchers and innovators to experiment with networks. new Internetwork architecture supports without the need extra mechanisms mobility, multi-homing Quality Service, provides secure configurable environment seamless adoption. implements core protocols multiple policies customize such different environments. Policies can be developed via Software Development Kit (SDK). an API applications natively use IPC...

In this work we discuss the limitations of link emulators based on conventional network stacks, and present our alternative architecture called TLEM, which is designed to address current high speed links be open future improvements. TLEM structured as a pipeline stages, implemented with separate threads limited interactions each other, so that performance can achieved. Our emulator handle bidirectional traffic at speeds over 18 Mpps (64 byte packets) 40 Gbit/s (1500 per direction even large...

Cloud providers are continuously evolving their virtual networking infrastructure to improve performance and functionality. This evolution, however, typically stops at the NIC interface, since any change in that domain would require impractical upgrades running VMs with collaboration of customers who own them. could hinder many important evolutions, like transition newer revisions VirtIO standard. To overcome this problem we propose BPFHV, a new paravirtualized network meta-device is able...

The assignment of available network bandwidth to application flows is a typical resource allocation problem. Although the nature this problem appears simple, inherent dynamism networks makes it hard achieve. TCP/IP protocol suite lacks capability flexibly reserve based on resources. We investigate flow with guaranteed Recursive InterNetwork Architecture, which allows deployment arbitrary strategies thanks its programmability. present distributed Flow Allocator policy that respects...

European funded research into the Recursive Inter-Network Architecture (RINA) started with IRATI, which developed an initial prototype implementation for OS/Linux. IRATI was quickly succeeded by PRISTINE project, different policies, each tailored to specific use cases. Both projects were development-driven, where most experimentation limited unit testing and smaller scale integration testing. In order assess viability of RINA as alternative current network technologies, larger experimental...

Effective capacity allocation is essential for a network to operate properly, providing predictable quality of service guarantees and avoiding bottlenecks. Achieving fairness long-standing problem extensively researched in the frame transport layer protocols such as TCP/IP. The Recursive InterNetwork Architecture offers programmable policies that enable more flexible control on mechanics flow allocation. In this paper, we present our version one these policies, which provides according...

The paper ‘A study of I/0 performance virtual machines’ by Giuseppe Lettieri, Vincenzo Maffione and Luigi Rizzo, published in Computer Journal: Section B - communications networks systems (Volume 61, Number 6), investigates some counterintuitive but frequent issues that arise when doing high-speed networking (or I/O general) with machines (VMs).