02 - The Design and Implementation of OpenFlow/SDN testbed RISE 3.0

Shuji ISHII(NICT), Tomoaki Takata (NICT), Eiji KAWAI(NICT), Yoshihiko KANAUMI (NICT), Shu-ichi SAITO(NICT), Shinji Shimojo(NICT / Osaka University)

This poster discusses how to virtualize the topology of testbed networks
for more flexible configuration of testing environment.

Since 2009, NICT have been working on developing a nation-wide OpenFlow testbed "RISE (Research Infrastructure for large-Scale network

The features of the RISE testbed are hardware-based packet transfer, to
meet those real-time and high-bandwidth applications, and SDN slicing
capability which provides users with their own OpenFlow environment. On
each slice, a user has her own controller, logical OpenFlow switches,
and virtual machines.

Through RISE operation, we faced the following two issues:
1) Poor capacity in terms of the number of concurrent users
2) Inflexible topology of underlay networks

In this poster, we describe the improvement of the RISE architecture "RISE 3.0" focusing on the second issue (inflexible topology). To solve this issue, we designed layered OpenFlow networks in RISE. The upper layer is for users and the lower layer is for logical (virtualized) topology administration, which works transparently to the upper layer.
Because those logical topologies, provided transparently to the upper layer (user) networks, consist of a wide variety of underlay (substrate) networks, we need a mechanism not relying on VLAN IDs. Thus, RISE 3.0 utilizes the MAC address fields, not the VLAN ID field, in the packet header to identify where the packet should be forwarded. With this mechanism, the users can request flexible network topologies in their slices, which highly enhances the capability of RISE as a wide-area OpenFlow testbed.

Download file