NS1's intelligent DNS and traffic management platform, with its data-driven architecture and unique Filter Chain™ routing engine, is purpose-built for the most demanding, mission-critical applications on the Internet. Features include:
- Anycasted DNS Network - 24 network POPs, 6 continents to ensure the best performance — any time, anywhere
- Point and Click Traffic Management
- Powerful API & Integrations
- Data Driven Delivery
In an office near the southern tip of Manhattan, the NS1 team is helping to direct Internet traffic just as the people three stories below are navigating the crowded city streets with Google Maps on their phones.
Here’s how it works: The Domain Name System, DNS for short, provides a worldwide directory service for the Internet. “When a user requests a site, most DNS providers make a guess as to where the user is located based on their IP and send them to the closest data center that’s up,” says Jonathan Sullivan, co-founder and chief product officer. “So it’s like GPS in the year 2000: You put in Home Depot, and it knows where the closest one is, but it has no idea what the traffic is like on the way there or what the conditions are like in the store—if it’s on fire, those kinds of things.”
Enter NS1. Just as Google Maps and Waze improved upon GPS’s capabilities, NS1 is doing the same for Internet traffic management. Pulsar, the company’s routing product, ingests telemetry from their customers’ applications as well as the Internet at large—not stores that are on fire, but rather things like server outages or heavy traffic on particular mobile networks—and uses the data to navigate around the potential speed bumps in order to optimize performance and resiliency for those sites. “It’s Waze for the Internet,” says Sullivan. “Everybody uses that analogy now. It’s kind of apt.”
Building a Startup
Now one of the “big three” DNS providers, NS1 was founded in 2013 by Sullivan, Kris Beevers (CEO) and Alex Vayl (VP, Demand Generation).
The three had worked together at Voxel, an early infrastructure and cloud provider. As part of the first cloud revolution in the 2000s, they started to see a lot of change in the ecosystem, with applications becoming much more distributed and dynamic. “We realized a lot of the ideas we had developed around managing traffic across distributed infrastructure at Voxel were applicable to everybody’s applications,” says Beevers.
Initially, the co-founders targeted businesses and startups that they felt were unable to get the right DNS technology in place. Early customers were primarily in the ad tech space, where every millisecond counts. NS1’s cutting edge platform was perfect for these kinds of demanding users, who were impressed by the performance and stability.
Business quickly boomed, but not in the direction the founders expected: while startups and agile “cloud native” users continued to drive signups, the business was inundated with demand from the biggest of the big—companies that NS1 had originally thought were fairly well served by entrenched providers like Dyn.
“Now our target customer is the top 20 percent of the internet—massive properties like Dropbox or LinkedIn that look to us for autonomy and redundancy at the DNS level, as well as unique performance optimizations that we can provide,” says Sullivan.
“Early on we focused on ‘power users’ with large traffic volumes and sophisticated setups,” says Beevers. “Now we find that our strongest demand is coming from traditional enterprises in the midst of cloud transformation, where our solutions help to make their mission-critical global infrastructure more resilient.”
Leading the Market with Three Products
Now NS1 offers three main products:
- Managed DNS—NS1’s global DNS network, which uses highly distributed infrastructure to deliver customer DNS traffic.
- Dedicated DNS—The same software platform, but deployed for customers either within their private environments or in dedicated cloud deployments.
- Pulsar—Traffic optimization using real-time telemetry from end users.
The use cases range from providing and directing traffic to a disaster recovery environment if a business experiences a data center failure, to routing traffic for large companies that have their own networks with dozens of data centers around the world.
Another use case that has become crucial in this age of frequent DDoS attacks is DNS redundancy. As Beevers explains, that means “having multiple DNS networks online that are physically independent of each other so that if one of them comes under attack, the other one picks up the slack and keeps their property online.”
we spin up and down servers at Packet, every day. Sometimes it is for capacity reasons, like to burst to absorb workloads, but often simply because we’re growing so fast and we need to scale up additional infrastructure.
In fact, “We’ve seen, especially in the last year or so, an alarming ramp up in malicious activity,” says Beevers. “We’re often the target of attack because of the critical position we have in our customers’ stacks, which means if we have a service-impacting issue, our customers have a service impacting issue, and their business is impacted. For us, that reinforces the importance of resiliency in our own infrastructure.”'
Scaling Capacity While Under Attack
This need for resiliency ultimately led NS1 to leverage Packet infrastructure. However, it didn’t start that way.
From the beginning, the NS1 team decided to place things like its APIs, databases, user interface and internal operational tools into a purpose-built data center environment, instead of opting for a cloud-based deployment like AWS or Google Cloud. Says Beevers: “We needed physical infrastructure around the globe to support our core workloads, and so we initially deployed in traditional colocation environments.”
But given the company’s growth—NS1 has grown several hundred percent per year, every year of its existence—the team felt that this static approach couldn’t keep up, and started looking for options to scale up on demand.
“Our very first use case for Packet actually was around mitigating large DDoS incidents,” says Beevers. “We were receiving a huge influx of malicious traffic and needed to burst our capacity quickly in key markets to absorb and mitigate attacks. Due to our need for high-performing bare metal servers, advanced networking capabilities, and low latency delivery, most cloud providers weren’t an option. Packet’s ability to provide these resources on demand enabled us to automate around our nuanced use cases.”
Now, Beevers says, “we spin up and down servers at Packet, every day. Sometimes it is for capacity reasons, like to burst to absorb workloads, but often simply because we’re growing so fast and we need to scale up additional infrastructure.”
Without Packet’s focus on automation, it simply wouldn’t work. NS1 has thousands of servers in use and a DevOps team of fewer than five people to manage them all. As such, Beevers adds, “the only way that our team’s scale is decoupled from our traffic and platform scale is through automation. Combining Packet’s rock solid automation with the predictability of bare metal infrastructure has been a huge win for us.”
Staying Ahead of Demand
NS1 measures both customer-facing data and internal telemetry at the rate of hundreds of thousands of data points per second, and all of that is coming back to processing and analytics software that it has deployed on a combination of colocated and Packet infrastructure. “Our customers are looking for better onboarding tools, more insights into their traffic, and advanced management tools, which means our core infrastructure is going to get more complex over the next year or two,” says Beevers.
The stability and agility of the NS1 platform—and that of their large blue chip customers—is tied to an ambitious and well-thought out infrastructure strategy. While the company still maintains many layers of redundancy, NS1 has put a lot of faith in their relationship with Packet. Why? In a word: focus.
“Packet is one of the most exciting substrate businesses that we’ve seen in a long time,” says Beevers. “Providing on-demand access to fundamental infrastructure is really critical for how a lot of applications are developing, including ours. It’s clear that AWS-style public clouds are not the answer for all things—there really are workloads that demand access to physical hardware. With Packet, we’ve found this perfect combination of bleeding-edge automation and access to raw hardware that enables us to do what we need to do.”
Sullivan puts it this way: “The old adage is, you can pick two of three when it comes to speed, reliability and cost. Packet actually delivers all three. You can get infrastructure in a click or an API call, it’s incredibly reliable because there’s no virtualization, and it’s cost competitive with the biggest players out there.”