Skip to main content

Hitachi

Corporate InformationResearch & Development

Flow Control for Higher Resiliency in Wide Area Distributed Cloud Computing

— Presentation at the 37th IEEE COMPSAC 2013, IWFIT —

August 8, 2013

Report from Presenter


Photo 1 Presentation scene

Yokohama Research Laboratory presented new paradigm of cloud computing and the key technology, titled "Flow control for higher resiliency in wide area distributed cloud computing" at the workshop of 37th Annual International Computer Software & Applications Conference (COMPSAC 2013) (Photo 1).

This conference was held in Kyoto, Japan, during July 22-26, 2013. COMPSAC is the IEEE Signature Conference on Computers, Software, and Applications. It is one of the major international forums for academia, industry, and government to discuss research results, advancements and future trends in computer and software technologies and applications. The theme of the 37th COMPSAC conference was The Expanding Sphere of Software and Data.

Summary of the Presentation


Fig. 1 Limits of conventional cloud
computing

Enlarge

The rise of big data has been fueling the ongoing traffic explosion, while the variety of applications has been diversifying the required communication qualities in cloud computing. The size of the "digital universe" in 2011 was 1.8 zetta-bytes. Unlike conventional applications such as e-mail, emerging M2M (Machine to Machine) applications require fast response and high reliability. One example of an M2M application is gathering the sensor data of machinery such as turbines, construction equipment, and pumps in different factories so that they can be maintained and remotely controlled.

Conventional cloud computing is burdened with rapidly increasing traffic that requires a fast response and high reliability (Fig. 1). Although terminals can connect to datacenters (DCs) from all over the world, some M2M applications are processed in only one or a few DCs. This means a DC is not always located physically near the terminals, which makes it difficult to quickly and reliably accommodate the increasing traffic.


Fig. 2 Use case of wide area distributed
cloud computing

Enlarge



Fig. 3 Proposed flow control architecture
Enlarge

To tackle this challenge, we have been developing wide area distributed cloud computing that improves the response time and reduces the amount of traffic flowing in the wide area networks (WANs) by processing and storing data at a micro DCs (µDCs) near the terminals. Various other distributed clouds have also been proposed (Fig. 2). However, it is difficult for the cloud to maintain service in the case of µDC failure or network congestion, for three primary reasons: 1) emerging M2M applications require higher reliability than conventional applications; 2) the robustness of µDCs is substantially lower than conventional DCs because µDCs lack redundant facilities; and 3) the µDC manager is not authorized to control wide area networks owned by carriers or ISPs.

We propose a flow control system that transfers flow to other µDCs by converting the L2, L3, and L4 headers of received packets via OpenFlow*1 when failure or congestion occurs. The proposed system enables flow to be transferred to other µDCs even from M2M terminals such as sensors that are unable to change the destination IP address. We also developed a distributed control system to solve the critical problem of switching delay caused by the packet-in mechanism used when applying the OpenFlow to the distributed cloud (Fig. 3).

There are three key contributions in this work: 1) flow control for higher resiliency in the distributed cloud by converting packet headers via OpenFlow, 2) distributed OpenFlow control architecture with centralized management functions (i.e., provisioning and monitoring), and 3) comparison of the architecture on the basis of performance evaluation.

(By YABUSAKI Hitoshi)

Related Papers

  • Hitoshi Yabusaki, et. al., "Flow Control for Higher Resiliency in Wide Area Distributed Cloud Computing," proceeding on 37th IEEE COMPSAC2013, IWFIT, July 2013.
*1
OpenFlow is a trademark of Open Networking Foundation.
  • Page top