These papers are made available as a means to ensure timely dissemination of scholarly and technical work
on a non-commercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders,
notwithstanding that they have offered their works here electronically. It is understood that all persons copying
this information will adhere to the terms and constraints invoked by each author's copyright. These works may not
be reposted without the explicit permission of the copyright holder.
NFVnice: Dynamic Backpressure and Scheduling for NFV Service Chains,
Sameer Kulkarni, Wei Zhang, Jinho Hwang, Shriram Rajagopalan, K. K. Ramankrishnan, Timothy Wood, Mayutan Arumaithurai, Xiaoming Fu, ACM SIGCOMM 2017,
Managing Network Function (NF) service chains requires careful system resource management. We propose NFVnice, a user space NF scheduling and service chain management framework to provide fair, efficient and dynamic resource scheduling capabilities on Network Function Virtualization (NFV) platforms.
The NFVnice framework monitors load on a service chain at high frequency (1000Hz) and employs backpressure to shed load early in the service chain, thereby preventing wasted work. Borrowing concepts such as rate proportional scheduling from hardware packet schedulers, CPU shares are computed by accounting for heterogeneous packet processing costs of NFs, I/O, and traffic arrival characteristics. By leveraging cgroups, a user space process scheduling abstraction exposed by the operating system, NFVnice is capable of controlling when network functions should be scheduled. NFVnice improves NF performance by complementing the capabilities of the OS scheduler but without requiring changes to the OS’s scheduling mechanisms. Our controlled experiments show that NFVnice provides the appropriate rate-cost proportional fair share of CPU to NFs and significantly improves NF performance (throughput and loss) by reducing wasted work across an NF chain, compared to using the default OS scheduler. NFVnice achieves this even for heterogeneous NFs with vastly different computational costs and for heterogeneous workloads.
PDF [1421.0 kB]
REARM: Renewable Energy Based Resilient Deployment of Virtual Network Functions,
Sameer Kulkarni, Mayutan Arumaithurai, K. K. Ramakrishnan, Xiaoming Fu, 26th European Conference on Networks and Communications (EuCNC 2017), Oulu, Finland,
IEEE, June 2017.
Network Function Virtualization (NFV) is becoming
more prevalent in Data Center, Telecommunication and
Enterprise networks, enabling the Virtual Network Functions
(VNFs) to fast replace the traditional dedicated hardware based
middleboxes. Ensuring high availability and fault tolerance of
VNFs is cardinal to meet the performance and service level
agreement requirements. Also, with the increasing electricity
demands in the Information and Communications Technology
(ICT) sector, especially for the data centers, the inclination
towards employing renewable (green) resources to power up the
data centers is also increasing. Mitigating the carbon footprint
and curbing the energy costs have been the driving factors for
push towards employing the green energy resources. However,
the Green energy supply is rather intermittent and unstable.
In this work, we study the impact of deploying VNFs in
Green Data Centers (GDCs) and make a case for addressing
the VNF reliability and high availability to effectively tackle the
stability concerns of GDC. To this extent, we present REARM,
which adopts the concept of Transient VNFs that rely on a
very short advance warning time to seamlessly migrate the
VNFs from GDC to a more reliable and stable Data Centers
(SDCs). Our experiments with container based VNFs demonstrate
that adaptive state transfer mechanism results into significant
reduction in both computation and communication overheads for
maintaining the NF replica, and warning time of 30ms is sufficient
to failover VNFs (serving 1K flows) within a data center to ensure
high availability of NFV services.
PDF [381.2 kB]
DRENCH: A semi-distributed resource management framework for NFV based service function chaining,
Argyrios G. Tasiopoulos, Sameer Kulkarni, Mayutan Arumaithurai, Ioannis Psaras, K. K. Ramakrishnan, Xiaoming Fu, , IFIP Networking Conference (IFIP Networking) and Workshops, 2017,
978-3-901882-94-4, June 2017.
As networks grow in scale and complexity, the use of Network Function Virtualization (NFV) and the ability to dynamically instantiate network function instances (NFls) allow us to scale out the network's capabilities in response to demand. At the same time, an increasing number of computing resources, deployed closer to users, as well as network equipment are now capable of performing general-purpose computation for NFV. However, NFV management in the presence of Service Function Chaining (SFC) for arbitrary topologies is a challenging task. In this work we argue for the necessity of an algorithmic resource managementframework that captures the involved tradeoffs of NFls minimum workload, load balancing, and flow path stretch. We introduce DRENCH as a low complexity NFV and flow steering management framework. In DRENCH an NFV market is considered where a centralised SDN controller acts as market orchestrator of NFV nodes. Through competition, NFV nodes make flow steering and NFl instantiation/consolidation decisions. DRENCH design enables third party NFV nodes participation while it can coexist with other NFV management solutions. DRENCH orchestrator parameterisation strikes the right balance between path stretch and NFl load balancing, resulting in significantly lower Flow Completion Times, up to 1Ox less, in some cases.
PDF [7212.9 kB]
Neo-NSH:Towards Scalable and Efficient Dynamic Service Function Chaining of Elastic Network Functions,
Sameer Kulkarni, Mayutan Arumaithurai, Xiaoming Fu, K. K. Ramakrishnan, 20th Conference on Innovations in Cloud, Internet and Networks (ICIN 2017), Paris, France,
IEEE, March 2017.
Middleboxes (Service Functions) have become indispensable part of Enterprise, Mobile and Data Center networks. Network operators rely on middleboxes to enforce network policies and to provide performance optimization, security and other value added services. With the increasing scale of network services and use-cases that require exercising specific sequence of service functions, the complexity and scale requirements of enforcing the network policies and orchestrating the service functions have significantly increased. Hence, dynamic service chaining and elastic scaling of network functions is fundamental for the performance of Data Center and Enterprise networks. To this end we propose Neo-NSH: an amendment to Network Service Headers (NSH), specifically to the service path header. We purpose the 24-bit Service Path Identifier (SPI) to express the service-chain ID instead of representing the service paths. We believe our proposal is more conservative and extends on the key benefits of the proposed NSH, by making it more efficient and scalable in facilitating the dynamic service function chaining of the elastic network functions.
PDF [286.2 kB]
Name enhanced SDN framework for service function chaining of elastic Network functions,
Sameer Kulkarni, Mayutan Arumaithurai, Argyrious G. Tasiopoulos, Yiaonis Psaras, K. K. Ramakrishnan, Xiaoming Fu, George Pavlou, IEEE INFOCOM Workshops 2016, pages 45-46,
IEEE, 978-1-4673-9955-5, SDN, NFV, Load Balancing, April 2016.
Middleboxes have become an integral part of Internet infrastructure, providing additional flow processing for policy control, security, and performance optimization. Network Function Virtualisation (NFV) proposes the deployment of software-based middleboxes on top of commercial off-the-shelf (COTS), enabling the dynamic adjustment of Virtual Network Functions (VNFs), both in terms of instance numbers and computational power. The performance of Data center and Enterprise networks depend strongly on efficient scaling of VNFs and the traffic load balance across VNF instances. To this end, we present Name enhanced SDN framework for service function chaining of elastic Network functions (NSN) that extends the Function-Centric Service Chaining (FCSC) with load balancing functionalities to achieve efficient network utilization while reducing the switch flow rules by 2-4x compared to traditional SDN approaches.
PDF [152.2 kB]