Senthil Arunachalam and the Art of Seeing What the Network Hides

Every office has a shadow inventory. A printer arrives on a Tuesday. A camera is moved after a renovation. A building controller joins the network with the social grace of a stranger slipping into a dinner party halfway through the soup. Soon there are hundreds, then thousands, of connected things. They perform useful work, emit traffic and accumulate risk. The official asset list, meanwhile, sits nearby looking confident and increasingly fictional.

This is the sort of mismatch that explains Senthil Arunachalam's career. His public record is not a parade of reinventions. It is a steady walk up the network stack: first the switches that move packets, then the wireless systems that let endpoints roam, then the software that identifies and protects everything connected. The titles changed. The underlying question did not. How do you understand a system whose complexity is always trying to outrun its map?

A career written in packets

Arunachalam studied electrical engineering at the National Institute of Technology in Tiruchirappalli, then earned a master's in the same field at California State University, Northridge. It was suitable preparation for an industry that makes software decisions feel physical. A line of code in a switch can decide whether a packet proceeds, waits or disappears. A wireless handoff can be seamless, or it can remind a user that the laws of radio have veto power.

At Cisco, Arunachalam held leadership roles in the Enterprise and Data Center Switching businesses. His ORDR biography credits him with leading development of the Catalyst 2K and 3K product lines, part of a switching franchise measured in billions of dollars. This is engineering at a peculiar scale. A defect is never merely a defect when the product beneath it sits in offices, campuses and wiring closets around the world. Reliability becomes less a virtue than a rent payment, due every day.

The patent record supplies the finer grain. One invention concerns source-specific multicast at Layer 2: getting traffic from a particular source to the right interested ports without turning the switch into a fountain. Another covers the efficient encapsulation of encrypted wireless station data between an access point and a switch. A third addresses fast handoff as a wireless client moves through a wired-wireless network. The titles are dry. The ambition is practical: make movement through a complicated system efficient, intelligible and dependable.

The résumé is less a ladder than a tour through the changing anatomy of the network.

When the endpoint stopped sitting still

Arunachalam later became Director of Engineering at Aruba Networks, responsible for software development across wireless and software-defined networking products. The move from switching to wireless sounds like a change of scenery. Technically, it is more like asking the scenery to move while the play continues. A client leaves one access point and approaches another. Its sessions should persist. Its identity should remain useful. The wired infrastructure underneath must cooperate without demanding applause.

The cast of collaborators is also revealing. Patent records from the Cisco years name engineers including Gnanaprakasam Pandian and Sheausong Yang. Those names reappear in ORDR's history: Pandian as chief executive and co-founder, Yang as chief scientist and co-founder. Technology companies often announce themselves as a new beginning. The people building them know that trust, like a protocol, is usually established through repeated successful exchanges.

The inventory that argues back

Arunachalam joined ORDR in 2015, the year the company was founded. He is described publicly as a founding member and head of engineering, and earlier company material used the title vice president of engineering. In 2017, Indian corporate records show him becoming a director of the subsidiary then called CloudPost Networks India. The older CloudPost name survives on his LinkedIn profile, a small fossil from the company's early life.

ORDR's problem space begins where spreadsheets lose their nerve. Conventional endpoint security expects software agents to be installed on conventional computers. Connected environments are less accommodating. Cameras, sensors, industrial controls and specialized appliances may not accept an agent. They may be owned by different departments, speak different protocols or remain in service far longer than anyone predicted. Discovery has to occur from the network, and classification has to derive meaning from evidence rather than a neat enrollment form.

The work asks for two temperaments that do not always share a desk. The first is a taxonomist's patience: observe a device, compare its behavior, decide what it is and maintain confidence as the evidence changes. The second is an operator's impatience: once risk becomes clear, do something useful. Feed the right context to existing security systems. Apply segmentation. Help a network-access-control program move faster. Make a policy enforceable without turning the environment into an archaeological dig.

Discovery is only the opening argument

The distinction matters because an inventory can be accurate and still be inert. A list tells an operator that a camera exists. A useful system must also establish where it communicates, whether that pattern has changed, which vulnerabilities bear on its risk and what controls the surrounding network can actually enforce. Each extra fact makes the answer more valuable. It also makes the engineering more demanding. Data collected at different moments must resolve to the same asset. Labels must remain stable enough for policy. Integrations must pass context onward without sanding away the details that made it useful.

Here Arunachalam's earlier work looks less like a prelude and more like the machinery beneath the present. Switches expose relationships between ports, addresses and flows. Wireless systems complicate those relationships as clients roam. Software-defined networking makes control more programmable, but programmatic control is only clever when its inputs deserve trust. The progression from observation to identity to action is therefore not a marketing funnel. It is a chain of engineering dependencies, and a weak link can make a perfectly attractive dashboard confidently wrong.

ORDR's language has evolved with the market, from IoT discovery to connected-asset intelligence and automated security. The practical burden underneath remains recognizable. An environment needs a living account of its devices, not a census taken last quarter. It needs behavior described over time, not a verdict built from a single packet. It needs policy that respects operational reality. Nobody wants a security control to improve a factory by stopping the factory.

This is where infrastructure engineers tend to earn their skepticism. Networks are excellent at exposing tidy ideas to untidy conditions: partial telemetry, old hardware, proprietary protocols, overlapping ownership and maintenance windows governed by the calendar of some entirely different department. A product that works there must be built around exceptions without becoming one. That task rewards the sort of experience Arunachalam accumulated across established platforms before bringing it to a younger company.

In 2020, Arunachalam shared ORDR's launch of an IoT Discovery Program and Ordr Core, adding a compact comment of his own: “We just launched.” It is not a manifesto. It is a builder's sentence, four words that place the emphasis on the shipped thing. His visible public activity follows a similar pattern: company releases, Cisco partnerships, connected-device security and the work around ORDR's products. The biography remains spare while the systems do the talking.

Context is the product

A 2024 USPTO assignment provides the latest technical marker in the record: Arunachalam transferred an application filed that April to ORDR. The public assignment identifies him as the inventor and the company as recipient. Two decades after the multicast filing, he was still putting his name on network-related intellectual property. Longevity in engineering can invite ceremony. The patent office prefers forms, signatures and dates.

There is a satisfying coherence to the arc. At Cisco, the central problem was forwarding. At Aruba, it was mobility and programmable infrastructure. At ORDR, it is context: recognizing the asset behind the traffic, understanding the behavior behind the asset and connecting that understanding to a control. Each stage depends on the last. Security may be sold in layers, but networks experience it all at once.

A policy can only govern what an organization can see, name and understand.

Arunachalam's story also offers a corrective to the mythology of the sudden founder. ORDR's engineering proposition did not arrive from nowhere in 2015. It drew on years spent with forwarding tables, roaming clients, encrypted tunnels, product lines and collaborators who knew one another's habits under pressure. Founding work often looks less like a lightning bolt and more like accumulated proximity to a problem.

The connected world keeps adding guests. Some announce themselves. Others slip in through a side door and begin sending packets. Arunachalam's career has been devoted to the useful hospitality of knowing who is in the room - and to the firmer discipline of deciding what each guest is allowed to do next.