OpenMUL steps up

OpenMUL found mention at Lightreading’s recent listing of popular open-source SDN Controller platforms. It has also featured as one of the top 10 trending projects at SDxCentral.

It is surprising to note that most of the other projects are heavily vendor driven but we are delivering all of this with a passion driven approach to solve real-life problems.

OpenMUL PRISM – connecting legacy world with SDN!

The major problem with legacy network equipment which spawned the SDN
movement has been the lack of simple and centralized management plane all the
while making the network highly agile. Various legacy devices like IP routers
had different and proprietary control plane to data plane interfaces and hence it
became impossible to centralize the control plane. Even the north bound
interfaces on top of control planes like SNMP, netconf varied wildly for each
device.

One of the major USP of Openflow has been standardized interface between
control and data plane. It gives us immense potential to centralize control planes
and provide centralized control to network operators.

Having such a separation is great but is it a good idea to reinvent the way
networks all around the world have been designed?

The simple answer-we don’t need to redesign the networks. Because networks
still need to talk, say, OSPFv2 or BGPv4. These are built into DNA of modern
network infrastructure. But, Openflow can be definitely be used to solve the
centralized management problems. Effort should be put to bring more devices
under centralized management umbrella while letting interface to the external
world unchanged i.e. use OSPF, BGP etc as is.

Read the whitepaper here

The open-source code will be available in upcoming release of OpenMul ‘Concave’ in Jan 2015.

Opendaylight Vs Openmul

Many SDN enthusiasts ask us what sets us apart from other bigger projects like OpendaylightODL is a great project no doubt and has many features but we feel ODL is not yet ready for prime time and its architecture is inherently prone to stability and performance problems related to switch and application handling. On a highly loaded and dynamic network, stability and performance are of prime importance and that is where Openmul delivers and wins. Check Openmul Vs ODL performance and find out more.

Apart from performance the major points which sets openmul apart from the competition:

1) ODL has Base Network Service Functions implemented inside single address space eg Topology manager, stats manager, switch manager, host tracker, ARP handler, REST server. It also means failure in any of these functions take the whole controller down. But that is not the case with Openmul which supports major base network functions as stand-alone and independent entities.
2) Direct integration with legacy protocols which means easy to integrate using MPLS-LDP, OSPF, BGP based inter-data center technologies.
2) Language is C/Python. Majority of networking developer community DNA is C  based. Can write base network service functions or applications either in C or python.
3) High flow download performance+latency with/without TLS. Most other controllers don’t support TLS.
5) Openmul provides the best flow, group and meter coherency between switch and controller. For example these entities are managed across switch, controller reboots/failover or when both switch or controller restart. Flows, groups & meters have strict applications ownership rather than controller itself.
6) Supports hot standby HA where controllers can run in active-standby mode. Automatic sync up of controller information between each other out-of-band.
7) OpenMUL can be easily customized. Core functionality, South-bound protocols and services can be added without affecting up-and running modules.
8)  OpenMUL has a highly distributed architecture. Base functions can be distributed/spread locally while one can use OpenMUL’s support for BGP to provide a geo-distributed controller abstraction.

Loop Detection feature of OpenMUL

This article explains the spanning tree implementation of MUL SDN Controller.

Introduction

This feature is used when there are one or more occurrences of loop in the network. Spanning Tree implementation makes sure that there are no loops when there are two or more paths to reach a particular network element. Once a loop is occurred due to a broadcast stream then it can be deadly for the network. So, this makes Loop Detection feature an antidote for any network.

loop-detection1

The CRUX

Loop detection module uses LLDP for implementing Spanning Tree.  When network element is added to a network, MUL SDN controller sends LLDP packets to every port of the attached network element. By doing this, MUL Controller gets the information about the network topology.

After getting all the information about the neighbours, port states needs to be decided.

Details of the process are :  Continue reading