XCLOUD CONDUCTOR
for Data Center Networks

Safely share single Data Center Network among customers or teams

Bringing Cloud-Based principles into Data Center switching fabric

Freedom of choosing various Hardware Vendors

General Overview

Changes delivered in minutes instead of days
Simplified and Unified architecture
Collaboration is improved

 XCloud Conductor provides single intuitive web interface which is used by all parties who needs anything related to the network infrastructure. Network engineers use it to maintain the network. System engineers use it to get access lists, server ports, load balancing, redundancy to be maintained. Management uses it to gather required insights. 

Here every network consumer can “use” the network the way it is required simply through the intuitive web portal, effectively avoiding ticketing mechanisms or other communication with network engineering teams. From supporting perspective – network engineering team is working on supporting the network effectively, avoiding time waste on manual implementation of tickets. This leads to improvements in the company lifecycle. Network change delivery time is improved from days into minutes. Collaboration is improved between different teams. Companies running own infrastructure can even share single deployment with their partner companies for additional cost optimization.

 Colocation or Cloud providers can safely share the network infrastructure with their customers. Doesn’t matter if single customer’s servers are located in different racks or different rooms or even different data centers. Inside XCloud Conductor platform, they all are connected to the single elastic fabric, grouped into every relevant tenant (customer).

 

Open Hardware, Operating System, Software

No sticking to single Hardware vendor anymore
Simplified and Unified architecture
up to 4 times more cost-effective

 White box or Bare-metal switches are blank, open, standard hardware available from various vendors. Based on standard silicon chips made by Broadcom or Mellanox providing low cost but industry standard wire speed switching with high density of 10/25/40/100 Gbps Ports

Those standard switches are available from various well-established vendors. Please check our hardware guide for the list of supported switches and technical specifications. Like open-source software once revolutionized server world – open-hardware is going to revolutionize the network infrastructure world. 

Cumulus Linux is a networking-focused operating system based on normal Linux which is running on physical switches.

XCloud switch driver is running on top of Cumulus Linux communicating with XCloud controller which is available as virtual machine image or even hosted in the cloud. Unlike orchestration systems, our software which connects a switch and cloud controller are working right inside the switch that means that adding more switches into existing fabric doesn’t add computational load to the orchestration system.

XCLOUD CONDUCTOR <br>for Data Center Networks

Architecture

Physical and Virtual servers share same approach
Borderless network & server mobility
Simple and scalable fabric
XCLOUD CONDUCTOR <br>for Data Center Networks
   XCloud Conductor Elastic Fabric is running BGP Unnumbered ( the extension which simplifies BGP configuration ). Extremely important detail is that BGP is used not only for a switch to switch communication but also for server to switch communication. Data Center switching solutions were emerging during decades eventually coming to very advanced fabrics which use Layer-3 technologies as transport for Layer-2 technologies. But server to network connectivity was still using old technologies like spanning tree, multi-chassis LAG, broadcast domain, network subnetting, interface bonding.

  With routing on the host (ROH) technology There is no need to transport Layer-2 shared segments, every server, and every switch communicate directly on Layer-3 – turning every server true part of the switching fabric and not something legacy which needs an overlay or any workaround. 

Some server virtualization platforms (currently Proxmox and OpenStack) are possible to seamlessly integrate with the XCloud Conductor platform. This allows both virtual and physical servers to use absolutely same and single approach to communicating with the network. Which is a crucial detail for creating agile and convergent deployments.

With BGP unnumbered configuration is very simple because there is no need to configure IP addresses on every link. Same, copied BGP configuration is used for all servers standardizing the configuration process.

Compatible with almost all Linux servers and MS Windows starting from 2012 R2.

Access Lists & Load Balancing

Centralized firewalls can be removed
Load Balancers can removed
ACL database & security procedures

Whether it is required to configure ACL, move server port configuration to another switch or configure new ports for connecting new server – doesn’t matter. No need to create a ticket and wait until network team will implement that manually. Now everything is done through the web portal.

The network can be isolated and assigned to tenants, then every network consumer can safely configure its part of virtualized hardware without single risk to affect neighbor’s services. System engineers can define Access Lists, Load Balancer features and migrate network side configuration between racks/switches right from the portal without a need to communicate with Network Engineering teams. 

Access Lists are defined through Portal. ACL approval procedures are optionally implemented inside the same tool – for cases when one team requires access to servers managed by another team or another company or if simply approval procedure is required by the security policy.

ACL rules are synchronized with the hardware efficiently in less than 1 minutes and unwanted traffic is blocked at every Ingress point directly by switches – without using separate firewalls. Our intelligent software applies ACLs the most efficient way to minimize usage of TCAM table of switching silicon.

BGP is used for redundancy and load balancing needs on Internet scale during decades. With our platform similar technique is used to organize traffic load balancing or server redundancy or redundancy of HAPROXY (or similar) in simple steps.

Health checks are running on every switch, actively testing every application server configured under load balancer in order to reroute traffic in case particular server is alive but the application failed.

This approach allows avoiding using separate load balancers in many deployments.

 

XCLOUD CONDUCTOR <br>for Data Center Networks