Tesuto is working towards an ambitious goal of eliminating network outages due to control plane operations, or software failures, within network devices. This effort requires building extremely large networks within public clouds in a timely manner. Not only do we have to build them quickly, we have to tear them down, modify them, and keep multiple copies of them! In the emulation of networks with thousands of devices, and tens of thousand of interconnections, the explosion of configuration and topology data is immediate. We have to know what interface on which device is connected to which interface on some other device. Tracking all of this information requires storage of state on millions of configuration elements simultaneously. In other words, Tesuto maybe a small company right now, but our compute and data storage needs are BIG.
You may be asking yourself is Tesuto truly scalable for my needs and ready for a network of my size? Maybe you are a tier one network handling ten percent of destination traffic on the Internet? We welcome you. Possibly you are an enterprise with thousands of critical routers around the world? Hi, let us help. We are designing and building our SaaS to support thousands of unique networks each containing tens of thousands of devices with thousands of interfaces per device.
Taking a step back for a minute, I would like to talk about what I have been working on before Tesuto. For the past several years I have been working on solving a singular problem as a principal engineer at Verizon Digital Media Services. The challenge I was tasked with was solving DDOS attacks against a global edge delivery network. With a global footprint of over 130 POPs and multi-terabit peak traffic loads, this was an interesting challenge. The goal, of course, for any content delivery network, is uptime and performance. The iterations of testing required to ensure optimal performance, during the biggest of attacks, were many. The solution ended up requiring analysis of billions of data points in near real-time. The result was a system that can identify and mitigate nearly all DDOS attack against the global network in sub-minute time frames.
Along the way Hossein Lotfi and I had the opportunity to work together before he went off to the SDN team at Google. While our focuses were different, our goals were the same: maximum network uptime and performance. Not that long ago Hossein and I were talking and sharing stories. He was telling me what it is like in Google, and we were commiserating over shared woes in the networking world. This is when he explained an idea he had to build an emulation environment for network changes in the cloud. It was an immediately interesting idea to me…although I wasn’t sure it was technically possible. What he was describing to me was the earliest framework for what is now Tesuto.
A lot has happened since that first conversation. Most recently, I have resigned from Verizon to officially start as Tesuto’s CTO. I am extremely excited for us to show you what we are up to. We have been busy preparing what we believe to be the most complete solution for network emulation and validation.
The design of our solution is a direct result of our years of working with some of the largest networks on the Internet. We are not relying on our own experience alone though; we have been talking to network operators and manufacturers about their needs when it comes to network validation and change acceptance. If you are someone with thoughts and opinions about this we would love to hear from you, email us at firstname.lastname@example.org. Questions we find ourselves asking network operators frequently these days are:
- Do the majority of your network outages occur during control plane operations?
- Have you performed lost revenue calculations on previous outages?
- Do you currently budget for network testing and validation?
- Do you currently have pre-deployment network change verification, whether it is a commercial product or something homegrown?
- Is their change resistance bias in your network teams due to lack of proper pre-change testing?
These questions are what are driving us to be the automated validation process for all network changes, both in Internet and enterprise networks. The methodology of continuous integration and continuous deployment has been a boon to software reliability and productivity, it is past time that the same functionality is applied to network device changes.
Tesuto’s solution allows for cloning networks with thousands of devices in minutes. Through our use of telemetry nodes we offer multiple vantage points into a network and jump boxes allow you to manually interact with your cloned network. Our validation framework can scale from a single test to thousands. We don’t just validate the end state of the network, we care about what happens during state transition, and provide tools to perform asynchronous testing, while device configuration changes are propagating through the network.
Whether you have ten mission critical devices, or ten thousand, we would love to chat with you. If you have interesting or exotic network requirements and wonder if our solution can help you please don’t hesitate to contact us. If you have plans to be at NANOG in Atlanta in February and would like to chat in person, please get in touch.