Can you guess the answer to this series of facts: what saved Facebook $2 billion, ditto
for Apple, cut Fidelity Investments’ data center electric bill by 20, put networking giant Cisco on notice, and convinced HP to stop fighting the movement and join it? The answer is the Open Compute project (OCP).
In the software world, open source has been revolutionary and disruptive. This movement created Linux, which is the software running most data centers around the world, and Android, the most popular smartphone platform out there today. Along the way, massively powerful companies like Microsoft, Nokia, and Blackberry were disrupted — some to the brink of extinction.
Now, OCP threatens to do the same to niche proprietary hardware based packet broker companies especially the ones whose business model is primarily based on selling custom hardware.
OCP’s impact on the NPB space
A network packet broker (NPB) is a device that provides a collection of monitoring tools with access to network traffic. NPBs make monitoring and security platforms more effective by feeding them data from the entire network stack. These products broker network traffic from multiple SPAN/TAP ports, manipulating this data to allow more efficient use of network tools and monitoring devices.
The Open Compute Project (OCP) is an organization that shares designs of data center products among companies, including Facebook, Intel, Google, Apple, Microsoft, Rackspace, Cisco, Goldman Sachs, and Fidelity. The OCP initiative was announced in 2011 with the mission to enable delivery of the most efficient server, storage, and data center hardware designs for scalable computing. In 2013, an effort to define an open network switch was announced to allow Facebook to load its own operating system software onto the switch. The hardware design was open sourced to enable cost effective solutions in other vertical markets.
Changing gears to Packet Brokers, today’s NPB architecture is based on proprietary hardware. It comprises hardware boxes you acquire from a vendor along with software, either embedded in it or licensed through what a vendor calls “smart licenses”. Let’s call this approach Packet Broker 1.0 architecture. However, as enterprises and service providers embrace new technologies, from virtualization to cloud computing and from the IoT to 5G, the focus shifts to making networks faster, flatter, and more efficient. Today’s networks must support ever-increasing traffic volumes, higher speeds and increased requirements for security, analytics, and compliance.
In order to provide the scale needed for these new and emerging services, NPBs need to decouple software from hardware. Software is where you innovate and scale, and to be more efficient NPB software must run directly on OCP switches.
The age of Packet Broker 2.0
It’s easier to do this now that merchant silicon vendors like Broadcom and Intel have ramped up the processing power of off-the-shelf chips for network hardware for Open Compute. The trend is similar to what happened in the compute market when x86 operating systems were decoupled from bare-metal servers. This strategy enables enterprises and service providers to take advantage of the latest advances in processor manufacturing technology at a lower price point, due to the prohibitive costs associated with the development and production of custom chips and proprietary hardware. Network Packet Broker (NPB) software running on Open Compute hardware is what we call packet Broker 2.0 architecture. It’s essentially a disaggregated architecture. With disaggregated hardware and software, engineers can design a highly available, non-blocking, and efficient architecture without the cost implications of a closed-source solution.
The easiest way to explain disaggregation is by example: In the server world, you buy a physical server and select the OS separately; organizations can use hardware from HP, IBM, Dell, Cisco, etc and then run software from Microsoft, Linux, etc. This is the way it’s been for a while in the server world.
However, in the Packet Broker world, if you buy a NPB from Vendor A, you run Vendor A’s NPB OS. The OS and hardware are integrated (i.e. aggregated), unlike the server world where they are separate (disaggregated). With Packet Broker 2.0 architecture leveraging Open Compute you can disaggregate the NPB software and buy hardware separately from OCP vendors at much lower cost points compared to a monolithic solution. When considering Packet Broker 2.0 solutions keep two things in mind:
- Stay away from controller based architectures as this makes the operational model complex
- Packet broker software should be able to interconnect OCP devices to one another to create a full mesh architecture that is self-aware and self-healing and even connect legacy packet broker 1.0 devices. This will protect investment you have made in the legacy 1.0 devices.
Open Compute, Open Benefits
Why would I want to do this? In other words, what benefit is there given that I’ve now taken responsibility to glue together the hardware and software? Indeed, it’s a radical change. This approach breaks long-standing packet broker constants and, at cursory glance, seems to complicate the support process.
So, why would anyone take this approach? Well, there are several benefits including:
- Significant price advantage (of the order of 50%) compared to proprietary NPB hardware
- Reduced vendor lock in, as you can now make hardware/software decisions independently and change out one or the other
- Simplified management
- Reduced maintenance costs
Perhaps most importantly you no longer need the expertise of Facebook, Google, or Amazon to experience the flexible nature of open networking. You can invest in a Packet Broker OS solution but retain control over what hardware to use. So next time you’re looking to buy hardware for a packet broker application think Open Compute and free yourself from the vagaries of vendor lock-in and high price points they unfairly command.