As time progresses, more and more innovative products are needed. The best example of the innovation at hand these days is the Intel®Tofino™ ASIC. With massive bandwidth and an increase in data availability, Intel®Tofino™ is just the thing you need. After all, who doesn't want flexibility and reliability in the DC Infrastructure?

Why don't you take a deeper dive into the Intel®Tofino™ technology? Here's everything you need to know!

Whitebox Switches: Fixed vs Programmable ASICS

ASICs are specifically designed for the special transmission of any kind of network traffic. Nowadays, two variants of ASICs are available; fixed and programmable, providing you with a number of options and possibilities.

The main difference between the two ASIC options is flexibility and adaptation. The fixed ASIC switch is the traditional approach which comes with a pre-defined data plane function structure and has a limited ability to manipulate the packet processing.

But what about the programmable ASIC switch? Are they the ones who can adapt to your needs by providing the flexibility required to establish your own or application-specific parameters? Definitely.

Programmable ASICs are the ones that offer methods to define the data plane functionality in a flexible way, using open source programming languages such as P4. This customizable pipeline or packet processing engine can be programmed according to your requirements. You can program the processing in innovative ways to make it compatible with any standard or non-standard Ethernet type. Furthermore, you benefit in getting the ability to build protocol specific extensions without the need to replace the switch ASIC by a new generation, as you would need to do it with a fixed function ASIC approach.

What is Intel®Tofino™ ASIC? Is it programmable?

What do programmable switches have to do with Intel®Tofino™ ? You will find that out in just a minute or two!

If you talk about Intel®Tofino™, it is a programmable switch ASIC that is based on a model called PISA (protocol-independent switch architecture) and allows you various opportunities that are otherwise not possible with fixed pipeline-based switches. At the same time, they provide you with full-fledged performance with your device so that it can reach its full potential.

Intel®Tofino™ provides you with bandwidth up to 400GbE speed per port today with a roadmap exceeding the current level by multiple factors. Diving into it deeper, it's important to understand that Intel®Tofino™ is programmable via the Open Source Programming Language called P4. This language was designed to program any kind of network device (switches, NICs, routers, etc.) and gives you the ability to control how packets are processed through your device.

Last but surely not least, to give you an interesting example, you are capable of defining your own process to identify bottlenecks within your network and tune it in an automatic AI fashion. You can also export flow-specific metadata information to external apps to in-depth analyze specific things.

Intel®Tofino™ – How It Started?

When you have a breakthrough technology such as Intel®Tofino™, you often wonder who were the genius people creating such an innovation. This revolutionary fabric processor was created and proposed by Barefoot Networks in 2013 under the co-founder's Nick McKeown, Martin Izzard, Pat Bosshart, and Stefanos Sidiropoulos. At that time, the ASIC was already running at up to 6.5 Tbit/s. After its massive success, it was purchased by none other than Intel® in 2019, where it got even more recognition, and this is what that technology deserves.

Data Center of the Future with Intel®Tofino™

Different organizations and different people propose a variety of structures for the data centers of the future. It is surely a topic worthy of debate, and each brings forward the framework that suits best for him.

However, according to Intel®, the data center of the future will be derived from Intel®Tofino™ and P4. They believe that the data center of the future will be many times more efficient and beneficial than they are now.

It can be explained by measuring the needs for the future. Present-day data centers are complicated, mixed up, have security issues, and have no way of creating an application-optimized forwarding with AI methods. All these complications will be solved by Intel®Tofino™'s ability to be programmed. Let's discuss a scenario in which you are using a fixed switch. You used it for VXLAN, but now you want IPL routing or MPLS. You cannot use a fixed switch as it does not come with much flexibility. But on the other hand, if you deploy a programmable switch, you can program each stage in different ways.

Intel®Tofino™ Generations

If you believe that Intel®Tofino™ has only one generation, you need to think again! Does innovation stop updating if it's successful? A big no!

Before Intel®Tofino™ was handed over to the Intel® group, it had already proposed the other generations of Intel, namely Intel®Tofino™ 1, Intel®Tofino™ 2 and Intel®Tofino™ 3. The third generation is soon to be published and launched in 2022.

Intel®Tofino™ 1 vs. Intel®Tofino™ 2 vs. Intel®Tofino™ 3

As for the three generations of Intel®Tofino™, it is understandable that they will have different strengths and weaknesses. On the other hand, they have different purposes and usability. The speed of the Intel®Tofino™ Generations is 25 Gbps, 56 Gbps, and 112 Gbps, respectively.

Intel®Tofino™ 1 supported only up to 100 Gb of port speed. But as for the other two generations, they can offer up to 400 Gb and level 2 schedulers. The only significant differences between the second and third-generation Intel®Tofino™ is the increase of resources, improved number of pipelines and a throughput bandwidth of 25.6 Tb.

Advantages of Intel®Tofino™

The advantages and uses of Intel®Tofino™ are immaculate. It´s hard to list all benefits the product can provide. However, some of the main benefits are worth highlighting. Real time in-band telemetry becomes a lot easier with the help of Intel®Tofino™. Adding metadata information to the packet processing creates a lot of options to monitor the network in-depth. Thus, opening many gates for more research.

One of the crucial parts of data centers is the load balancers. They are used in almost every data center, and they regulate the traffic that comes into data centers. However, Intel®Tofino™ can exclude the need for load balancers from the data centers altogether by acting as load balancers themselves.

The greatest benefit of Intel®Tofino™ ASIC switches is that they open the gates for many other opportunities. It acts as a platform for many other options and applications. Moreover, they are designed in a way that they require less power and space, making them the perfect switches to deploy. It can be safely said that Intel®Tofino™ has great potential to evolve in the future and provide endless possibilities with a little more effort and time.

If you have any more questions, don't hesitate to drop us a message. We'll be sure to contact you as soon as we can! After all, your contentment is our greatest responsibility.

Want to Learn More About it?

Get in touch with us and we will answer all of your questions.

Contact Us

Edgecore Networks Corporation announces general availability of the Wedge100BF switches with a programmable Tofino ASIC for high-performance data center network environments. The bare-metal top-of-rack (TOR) or spine switches support 32x100G ports, enable greater flexibility, control and innovation of enhanced network services, and support the installation of various open and commercial network operating system software.

STORDIS is an official partner of Edgecore Networks and your reliable source to obtain versatile Edgecore devices. Contact us for more information about the Wedge100BF series.

We offer special rates for members of academia and research institutions.

The data center market is undoubtedly changing towards disaggregated network infrastructures and industry experts recognize a trend that has been rapidly evolving over the last few years. Even legacy vendors have started to integrate third-party commodity silicon into their proprietary solutions to meet the operators’ needs for acceleration and uniformity. The article ‘Why Merchant Silicon Is Taking Over the Data Center Network Market’ by Emily Carr, published by Data Center Knowledge, talks about the increasing relevance of merchant silicon for today’s data center infrastructures.

IHS Markit analysts expect that 63 percent of all Ethernet switches shipping in 2022 will be merchant silicon, and even legacy vendors such as Cisco and Juniper are following this trend for disaggregation. Mike Bushong, VP of enterprise and cloud marketing for Juniper, estimates that the adoption of merchant silicon will increase rapidly throughout the data center industry. Not only are the benefits of economic nature, but operations can also be performed faster and much more efficiently by reducing diversity.

A decisive circumstance fueling this development occurred in 2016, when companies such as AT&T and the Chinese hyperscaler Tencent employed Tofino-based Barefoot Networks switches with a P4 programmable chipset. By incorporating open source software to their proprietary solutions, legacy vendors like Arista and Cisco eliminate the need for technologically advanced programmers to benefit from the complexity of white boxes, making their solutions accessible to companies of any size.

Alexander Jeffries, founder and CEO of STORDIS, talks to Data Center Knowledge about his experiences and what role the ‘STORDIS Advance Programmable Switches’ (APS) are playing in this development. Jeffries has witnessed the transition of data center networks for more than ten years, he knows the industry, he talks to customers and understands their demand for certain kind of hardware devices.

‘The kind of customers we’re talking to for these kinds of switches are quite a bit different to what we were talking to in the past,’ Jeffries says. ‘We’re getting a lot of interest from service providers – so telecommunication companies – the security environments, lots of interest in academic research, from all kinds of use cases,’ quotes the article.


Jeffries expects that open source will become standard in the network industry and supports the acceptance and use of open source in the German market. As a result, STORDIS is an active member and contributor to open source projects and organizations such as ONF (Open Network Foundation), P4 Language Consortium, and Open Source Business Alliance.

The company designed the APS to provide P4 programmable, open, high-performance bare metal switches delivering versatile features, such as INT and PTP, in a unique combination with powerful hardware.



The original article ‘Why Merchant Silicon Is Taking Over the Data Center Network Market’ by Emily Carr was published by the leading online source Data Center Knowledge on June 5th. We are not liable for the content of or any opinions expressed in the original article. Alexander Jeffries, CEO of STORDIS gave an interview in the context of the OCP, which was included in the article.

You can read the full article at

Aaron Glenn, network enthusiast and managing director of AAGICo Berlin UG, talks to Ethan Banks and Drew Conry-Murray of Packet Pushers about the innovative and beneficial deployment of P4 within Open Networking infrastructures. Aaron gives a profound, in-depth insight into the programming language, explains in detail how P4 works, and presents use cases.

P4 (programming protocol-independent packet processors) is a domain-specific language which is used to specify the packet behavior and unambiguously define the forwarding plane regardless of the underlying hardware, while the program defines the packet headers and the processing logic. It enables the reconfiguration of parsing and the processing, while the hardware (target), which can be an ASIC, an FPGA or a NIC, acts as a compiler, providing information and instructions via match-action tables.

Although P4 might appear to be a general-purpose programming language, it is not; neither is it a successor of OpenFlow. In general, it doesn’t provide the abstraction to create packets due to its boundaries to computation and memory consumption, and it only specifically enables the user to modify the packets on a very low level (from Layer 2 to Layer 4). It can represent any packet forwarding behavior, not only for Barefoot Tofino, which is also the first hardware switching ASIC that natively supports P4.

In the presentation, Aaron explains the program and architectures, the main components of a P4 program, and how it functions within the network environment. He illustrates the essential elements of P4 and various use cases of P4Runtime, the control plane interface to P4 for fixed-functions devices. He conlcuded with how to get access to P4 and provides valuable web links to online tutorials.



Packet Pushers

Packet Pushers is an independent community of IT practitioners and vendors who provide focused, profound and practical content covering networking, data center design, virtualization, and general technology trends. The community is primarily built around a successful podcast (up to 250K downloads per month), but they also operate a blog and a forum to share practical and technical information and evaluation. A major part of the work relies on sponsorship, enabling the members to perform essential research and ensure valuable and high-grade content.

STORDIS Advanced Programmable Switches with Barefoot Tofino ASIC

If you would like to learn more about the beneficial use of P4, the Tofino-based STORDIS APS (Advanced Programmable Switches) and how they can enhance your own enterprise network, you are very welcome to visit us at or contact us directly at or +49 (0) 711 34 21 58 0.

Disclaimer: STORDIS was not involved in the production of the webinar with Aaron Glenn. The webinar was published initially by Packet Pushers at on March 25th 2019.

Interesting article about Barefoot Networks on SDxCentral.

“One of Barefoot’s most-touted capabilities is that it enables programmability of the forwarding plane. “With the programmable chip what you can actually do is assign the resources to only the pieces you want (…)”, said Sylvanus “Prem” Jonnalagadda”. ” Tofino could also replace traditional load balancing. Layer 4 load balancing is typically done by appliances that are very expensive. But with Tofino that functionality can be embedded in the switch itself instead of having an expensive middle box.”

Would you like to know more about Barefoot, Tofino or the P4 Language Consortium ( Then get in touch with us.

Make sure to follow STORDIS on LinkedIn for news and updates about Open Networking.