Information Technology Reference
In-Depth Information
data center (DC) or an enterprise network or an AS (Autonomous System). The main
contributions of this paper: (1) we propose that different domains can run heterogene-
ous NOSes at the same time. (2) We define what network view information can be
exchanged among heterogeneous SDN peers. (3) We propose to abstract a physical
network into a virtual network for privacy or scalability. (4) We design a high per-
formance network view exchange mechanism. We refer to this mechanism as the
East-West Bridge (EWBridge for short).
The rest of the paper is organized as follows: Section 2 presents the related work.
Section 3 describes the design of EWBridge among SDN peers. Our work is imple-
mented and evaluated in Section 4. Section 5 concludes the paper.
2
Related Work
Currently, there are mainly two kinds of NOSes: (1) Single NOS, (2) Distributed
NOS. Single NOS is such as Floodlight [18], NOX [20], Maestro [21], Beacon [4],
SNAC [22], and Trema [10]. Distributed NOS is such as Onix [7], HyperFlow [8],
and DIFANE [17].
In the SDN centralized control model, all the routes are determined by the NOS, so
the first packet of each data flow is sent to the central NOS. Then, the NOS will com-
pute a routing path for each data flow and install the routing path into the related
OpenFlow [9] switches according to a global network view. Here, the first packet of
each data flow is usually called the “packet-in”. To improve the flow initialization
requests setting up speed, some NOSes such as Beacon and Maestro are trying to
improve the performance of the controller by the multi-thread technology. The NOS
software systems are always being deployed on a multicore host or server. However,
for large-scale data centers or networks, the request processing capability of a single
controller is limited: (1) NOX could process about 30K requests [23] per second; (2)
Maestro could process about 600K requests per second.
In fact, large-scale network environments always have vast amounts of data flows:
(1) a 1500 server cluster might generate 100K requests per second [24]; (2) a 100
switch data center might generate 10000K requests per second [21]. To achieve a
scalable control plane, distributed NOSes are proposed. Then, the scalability in the
control plane evolves into a cooperative performance problem of multi-NOSes. Such,
the east-west interface and communication has been considered as a very import
module in software defined network architecture, such as SDNi [5] and SDN at
Google [6].
HyperFlow [8] uses multiple controllers to construct a distributed control plane and
each controller takes charge for a small area of network. To learn a global network
view, HyperFlow adopts a distributed file system named WheelFS [15] which is de-
signed for the WAN (Wide Area Network) environment. Each HyperFlow controller
has the right to deal with network events within a certain local area, and the events
which will affect the global network should be announced from time to time. Once
other controllers learn the event information, they should replay the event to achieve
the synchronization of the global view. This approach can only deal with events
