Two major trends are happening in the world of data right now — one a direct result of the other.
A surge of big data and IoT connected devices, which software developer Stateless estimates will surpass 50 zettabytes and 51 billion devices respectively this year, has forced businesses to rethink how they can use networks to take advantage of all of the new information.
Many have ditched their on-premise network infrastructure in favor of options that offer better performance, security and scalability. According to INAP’s annual State of IT Infrastructure Management Report, nine out of 10 companies with on-premise data centers plan to relocate some, if not all, of their workloads off-site within the next three years.
So where are these organizations turning for the infrastructure tools they need?
Stateless Engineering Manager Ali Khatami said most combine cloud services like AWS or Azure with colocation data centers that rent out data center services and bare-metal servers, which are physical servers dedicated to single tenants. By combining several options, they’re creating hybrid-cloud environments.
Stateless says it supports these service combos through its consolidated software platform that makes network connections simple, elastic and scalable, accomplished using a patented software architecture that separates state from processing.
“It’s so unique for us to be in the state of Colorado actually doing this type of work,” Khatami said. “Historically, Colorado has been known for running data centers and storage on this type of traditional networking; our types of startups generally exist on the West Coast.”
By their “types of startups,” Khatami is referring to those dedicated to re-configuring networking infrastructures to the cloud, in a move to optimize control, connectivity, access and security over existing network setups.
Stateless addresses this through its patented technology used in its microservice-based Luxon platform. The platform provides data center operators, cloud-managed service providers, communication carriers and businesses with fast connections, remote access, interconnectivity and the ability to deploy their own network custom applications.
“No one is doing directly what we’re doing,” Khatami said.
As they explore what Khatami considers unchartered territory, Stateless engineers have come across complex challenges, opportunities to be heard, and plenty of new and emerging tech to help them navigate this new — and niche — industry.
One of the bigger focuses on Khatami’s team right now is unit testing of the component tree and all of the ancillary services that are used on Stateless’s platform. Their test environment includes local development workstations and running unit and integration tests using tools like Storybook and Jenkins.
Tamayo joined as Stateless’s third engineer and works on the “bootstrapper,” which is the method of initiating Stateless’s operating system. She collaborates frequently with other teams, including DevOps and orchestration, to come up with the most efficient and user-friendly design possible.
Kaplan joined Stateless in January from a high-frequency trading role. For Stateless, he works on the NFAPP team to build secure communication links for networks. He said he often hosts pair programming sessions with anyone who works on system components that are unfamiliar to him, allowing the team to quickly resolve difficult problems.
The state of Stateless
Corporate giants like Juniper Networks, Cisco and Palo Alto Networks dominate traditional, land-based network setups that require expensive and specific hardware configurations, Khatami said.
Unlike those network providers, engineers at Stateless are breaking virtual and hardware monoliths solutions into microservices and decoupling their state. Stateless’s Luxon software runs on these commodity servers and can be deployed in business, enterprise, data center or service provider environments.
The company hopes to transform legacy, monolithic networks that can take weeks to connect new digital resources by offering a microservice network function platform that’s flexible with changes, new services and scaling, and provides connectivity to nearly anything at the click of a mouse.
Khatami: We are a software company, but there is no hiding that we have to run on hardware to accomplish our tasks. We’re trying to lead a shift away from specific devices that only do a very specific thing to a lightweight more flexible piece of hardware that runs on performance software. Our first objective is to connect hardware to hardware, or data centers to cloud providers, to accelerate the adoption of hybrid cloud. Our cluster is designed to easily compose these very complex networks and make it easy to configure API-driven or UI-driven approaches.
Kaplan: Designing core networking infrastructure that is powerful enough to handle internet-scale traffic is a complex job. We often pair and mob program with other teams to narrow down system problems, and write regression tests to assure the problem doesn’t crop again. Tinkering with kernel and other low-level system components often requires both breadth and depth that only collaboration can provide.
Connecting the core
Four core engineering teams contribute to the ongoing development of Stateless’s platform.
The orchestration team leverages Docker containerization orchestration software to ensure the cluster operates as it should. They also maintain Stateless’s open API specifications so that customers and other teams at Stateless can integrate with the API.
Stateless’s NFAPP team, which Kaplan sits on, builds network functions like routers, packet filters, network address translation services and secure communications. He and his teammates look into how operating systems scheduler works, what types of optimizations can be made on a low latency problem set, and how to bypass operating systems to make networking processing faster.
Engineer Maria Tamayo works alongside the front-end team to create architectures that support clients with and without their own user interfaces. Her team is also designing what is referred to as the “cluster creation guide,” which is how customers set up their clusters from scratch.
Last but not least is Stateless’s DevOps team that maintains their CI/CD environment, which runs on the open-source automation server, Jenkins.
While the teams’ responsibilities are split, collaboration and communication across all four run rampant.
Tamayo: We have strong communication on all of our teams. Before every sprint, we hold a discussion on what issues are most important for us to prioritize. When a challenge comes along, we address it through communication.
Khatami: We try to run as Agile as possible, and we’re using JIRA and other Agile software to facilitate that work. When discussing architectural decisions, the team comes together to talk through the impact on the product, the value to the product, the difficulty of implementation and ease of implementation. We have a well thought-out process for choosing new architectural design changes.
Seniority takes a seat
Stateless’s open environment on the engineering team holds no biases when it comes to seniority.
In his second week, Kaplan suggested to his manager the practice of declarative programming on state machines to make reading, testing and logging code easier and faster.
Instead of getting stonewalled as he had at previous companies, Kaplan said his manager encouraged him to put together a presentation and present it to the team. Tamayo said she’s felt similar support.
Inspiration for this can come from anyone.”
Tamayo: I suggested a while back that every time we were going to make a change to our process, we list out together the pros and cons. We all have our own opinions, and they are all valid, but doing a list of pros and cons creates a logical, objective process in decision-making that keeps pride and passion out of it. That has helped us a lot; we actually just did a pros and cons list for managing events in our application.
Khatami: Our overall goal is to change the way networking exists. Inspiration for this can come from anyone, and the product can be moved by any person on the team, regardless of their seniority. We would be doing ourselves a disservice if we didn’t listen to everyone.