OrionVM Builds Breakthrough IaaS Cloud Platform with Cumulus® Linux®
Industry
Cloud Computing
Business Objective
Expand their data center
Technology
Edge-Core | Agema
The Situation
In our search for an open network solution, we needed an agile, cost-effective vehicle to expand our data center,” said Sheng Yeo, CEO of OrionVM. “Cumulus Linux fit the bill. We can now hit the ground running with the flexibility to manage our networks like any other server in the data center, while leveraging existing Linux network management daemons. We love the model and approach, and have it deployed in production already.”
Leverage world-class disruptive/new technologies for its breakthrough wholesale cloud platform, OrionVM needed a better approach to networking. Additionally, they needed to consider the following: Price performance with a 10G highspeed network, automation, and simplicity to enable rapid network deployment and innovations.
- Build wholesale cloud product suite with best-of-breed networking technologies
- Evolve networking into a truly automated process; break away from vendor lock-in
- Add platform features and functionality while preserving resources and maintaining top price-to-performance ratios
The Solution
OrionVM’s actual deployment consists of servers running Linux and the Xen hypervisor. OrionVM has built a custom cloud platform utilizing InfiniBand. The OrionVM cloud connects to a network infrastructure running Cumulus Linux on top of Edge-Core and Agema networking switches for cross-connect.
The network architecture is a two-tier architecture with an L3 Fabric, which is a combination of OSPFv2 and BGPv4 (Quagga routing suite). The Prescriptive Topology Module daemon (PTMd) helps with connectivity verification against a topology graph. Alongside Cumulus Linux, OrionVM leverages existing Linux daemons for L3 gateway failover, and its own daemons for orchestration and monitoring/troubleshooting. The open nature of Cumulus Linux helps create differentiation. OrionVM’s team can program the network for optimizations based time of the day, service level agreements, and other parameters driving price/performance efficiencies.
The Results
The CAPEX savings with the Cumulus Networks model enabled a faster fabric comprising wire-rate 10G switches. This was a ten-fold performance increase at the same cost! OrionVM also realized significant OPEX savings. The same engineers that managed the hypervisor Linux networking stack can now manage the switches. The network architects can focus on the architecture, and the overall team can gain more widespread control and easily develop new platforms/services. All this leads to team efficiencies, and servicing customers with faster speed of delivery.
Cumulus Linux provides the price/performance, and open ecosystem needed to achieve OrionVM’s goal of high performance, agile infrastructure at a fraction of the cost of others. Customers can get rapid service delivery with a cloud infrastructure available within minutes. OrionVM was able to deliver the same model proven in its Linux compute environment in our networking environment.
- Leverage same compute management infrastructure on network switches. Ability to leverage the same team for multiple workflows – Lower OPEX
- Rapid service delivery to end users by leveraging the same existing Linux orchestration automation and management infrastructure
- Ability to deliver higher bandwidth capacities to end user at same pricepoint 1G – 10G
The OrionVM internal stack is primarily Linux, so having Linux on the networking gear was a natural fit. Moreover, Cumulus Linux breaks away from locked in and proprietary CLI – Cumulus Linux is Linux after all, and what’s even more important is its open ecosystem with the following advantages:
- Daemons that run on servers can now run seamlessly on switches
- The same automation stack just works on networks without much integration
- The same tools used for Linux routing suites in the past can easily be leveraged for Cumulus Linux
- Monitoring and troubleshooting can be done via existing Linux tools and it’s easy to build on top of those tools
Overall, having Linux on networking gear simplifies data center operations.
“Cumulus Linux enables you to do more with current resources at 1/3 the initial price of the competition. And rapid service delivery means that where it used to take 24 hours, customers can now get their own network ready within 10 minutes,” said Sheng Yeo of OrionVM.
What’s next? The next generation platforms supported by Cumulus Linux 2.0 provide even more flexibility with a bigger buffer, larger flexible routing tables, richer ACL counters, and network virtualization. Flexible L3 routing tables can help effectively scale connectivity. Richer ACL counters could serve to track routing optimization. And the current cluster runs its own modified version of SDN overlays, so hardware SDN technologies could then be leveraged for SDN overlays over the existing network infrastructure.
