The first, programmable data plane platform in the Linux Stack

KulCloud PRISM is migrating network hardware to Linux

PRISM Overview


Kulcloud PRISM is a software approach to abstract legacy L2 and IP routing/forwarding using Openflow and Linux. It lets centralized logical routing instances control any set of geo-distributed or local data planes to better utilize resources, improve scalability and simplify management. Its dis-aggregated architecture is an enabler of agility and elasticity, which are key requirements for cloud deployments and applications. Kulcloud PRISM’s abstracts Linux on top of any supported hardware and offloads forwarding to the hardware from the Linux kernel.

Because, it’s just Linux, Kulcloud PRISM provides the ultimate flexibility for modern networking design and operation, enabling the best of breed hardware and solutions ecosystem with a standard operating system. Customers are able to manage whole network hardware in the single interface leveraging standard server automation & monitoring software and innovate rapidly with custom, open-source or commercial tools.

PRISM is open bridge between hardware box and linux applications

Kulcloud PRISM migrates a network hardware to Linux deploy every linux applications to hardware box


Choice of hardware, software, linux environment, customization, no vendor lock-in


Automated discovery of network and equipment plug & play of 3rd application


Supports control plane HA(Master/Slave) & fast fail-over on data plane


Switch cluster act as complete abstraction of single linux server

Use Case

PRISM supports on-demand virtual router instances

Flexible legacy routing protocol stack integration(Quagga, ZebOS.. Etc)

PRISM supports a various network plugins of cloud orchestration system

PRISM supports Openstack L3 service plugin & Docker network plugin

PRISM automation provision, updates and decommissions network and service

PRISM supports a seamless integration with legacy monitoring system

PRISM can reuse legacy monitoring system
PRISM can supports every legacy linux monitoring agent
SNMP, SSH, etc



Core FeaturesDedicated L2/L3 tables
Flow download modeReactive/Proactive
Max Cluster size16
Max portsAs Supported in switch
Cluster backplaneNon-blocking 8-way ECMP

L2/L3 Features
IPv4 Routing
No. of Routes
No. or Next-hops
Layer2 Support
Static LSPs, RSVP/LDP (Q1 2017)

DataCenter Gateway Features
No. of Tenants
No. of Networks
Namespace DHCP
Floating IP s
Per Tenant/Network Bandwidth profile
No. of hosts
Distributed Routing
DVR Supported
Distributed VTEP
To be supported(Q1 2017)
Integrated FW/LB
Available from supported vendors

Compatible Switches/OS
Pica8 (DC/Enterprise)
OpenVswitch 2.3.x
HP 38xx
Noviflow (Telco PE/Service Chaining)


PRISM CONX for IP Fabric

Network fabrics are defined as high-performance, low-latency, scalable ethernet switching products that connect compute, storage and software elements in a converged network. PRISM CONX enables any-to-any connectivity between nodes on the network and allowing multiple switches to be managed as one. The advantages of this model include the ability to offer application programming interfaces that directly link applications to the network fabric, thus improving Quality of Service and reducing latency

PRISM Application

PRISM application is the SDN compiler application to translate a linux network stack informations to white box switches using Openflow protocol.  It means that whole linux network application can run on white box switches and developer don’t need to know about Openflow protocols


PRISM agent is the event notifier of Linux network stack to PRISM application. If user install this agent on there linux and connect with PRISM application, they can manage directly every  route tables, ARP tables, ACLs and interface management of white box switches on remotely.


PRISM is integrated deeply with popular cloud management systems such as OpenStack  and can easily be integrated to nearly any environment using our APIs. By integrating with these platforms, PRISM makes building out and scaling networks a cinch.



PRISM can be centrally configured, or even completely automated via our RESTful API.


We’ve got a fancy command line interface just for you! It provides auto-completion, and a bunch of tools to help you manage your network deployment.


1 Where can one find the logs of PRISM?

It can be found along with system log files usually placed in /var/log/syslog.

The specific files are /var/log/syslog/prismapp* & /var/log/syslog/prismagent*

2 PRISM software stopped working. What to do?

Make sure your license period is active.

Check log files to find any errors. Else, contact Kulcloud support.

3 The logs say license expired. What to do?

Make sure your license period is active.

If so, make sure system clock of the installed system is correctly setup. If nothing works contact Kulcloud support.

4 What routing protocols does PRISM support?

PRISM currently supports routing protocols supported by default quagga software.
KulCloud also supports hardened version of quagga. Please contact KulCloud sales for
the same.

5 What kind of HA is supported by PRISM?

PRISM currently supports warm standby HA from a routing protocol perspective. The
SDN controller however operates in hot standby mode and is able to immediately take
over proceedings once a HA state transition is detected.

6 Does PRISM support LAG interfaces?

Currently, as of ellipse-release r5.3, PRISM supports static LAG channels. It is to be
noted that PRISM does not require traditional LAG internally in the fabric as it leverages
Openflow groups to implement ECMP.

7 What are the scaling numbers expected from PRISM?

Please refer to PRISM data-sheet for more information or get in touch with KulCloud support.

8 Can PRISM support head-less mode where PRISM software is dead but forwarding continues to take place?

PRISM supports headless mode but once ARP caches in neighbours or hosts expire
there will be an outage in traffic forwarding since ARP is handled by PRISM control plane.

9 When PRISM is running in HA mode, is it possible to provision in slave node also?

No, it is not possible to provision in a slave node.

One can check the HA state of a particular node and proceed with provisioning.

10 What is the downtime in forwarding during HA failover?

There is a very slight downtime due to convergence time of the routing protocols.

11 How does PRISM handle multi-tenancy?

Multi-tenancy is handled by PRISM by leveraging Linux namespaces.

Provisioning can be done on a per-tenant basis.

12 Are there any scalability concerns due to namespace networking for multi-tenancy?

Namespaces provide solid tenant isolation and with per-process resource utilization in
Linux being quite low, one should be easily be able to scale more than 4000 tenants in
a single box. Furthermore, PRISM is a based on distributed architecture and can be
easily scale-out beyond a single box.

13 What kind of hardware/switch dependency does PRISM have?

As of ellipse release r5.3, PRISM supports PicOS-2.6 (from Pica8), Noviflow switches
as well as Openvswitch 2.4 (or higher). Depending on switch used, performance and
scalability number will vary but the PRISM provisioning remains the same.

14 Does PRISM support distributed VxLAN termination ?

As of ellipse release r5.3, PRISM supports single switch VxLAN termination and distributed termination would soon be available.

15 What is the turnaround time if users request an urgent feature to be
included in the PRISM software?

Normally we do our best to support all the requests from all customers but each request
has to be evaluated by development team before we can commit to a release schedule
for the same.


If you have another issue or problem, you can report by email.

email :