Over 7,000 devices connected simultaneously via Wi-Fi. 300 wired clients. Over 200 next-generation access points. A highly redundant 10 Gbps backbone. And all of this installed, configured, and operated in one of Spain's largest exhibition centers.
This was the connectivity deployment Enbex executed for the European edition of the World Linux Congress, held at Fira de Valencia.
This article is not a commercial summary. It is a detailed technical tour of each layer of the project: from network infrastructure design to real-time management of thousands of simultaneous connections, including the specific challenges of serving an audience made up entirely of tech professionals. In other words, the type of user who immediately notices if something isn't working as it should.
The context: What does it mean to provide WiFi for a world technology congress
Before diving into the technical aspects, it's worth understanding the context. The Linux World Congress, in its European edition, brings together thousands of developers, system engineers, infrastructure managers, and open-source community leaders from around the world. It's an event organized under the umbrella of the Linux Foundation, the organization that supports projects that power the internet as we know it: Linux, Kubernetes, Node.js, among many others.
The venue was Fira de Valencia, a very large exhibition center with multiple halls, conference rooms, exhibition areas, and common spaces. The challenge wasn't simply «putting WiFi in a big place.» It was about providing high-performance, stable, and secure connectivity to a massive audience that is also extraordinarily demanding about network quality. We're talking about people who develop software, deploy cloud infrastructure, and debug network issues as part of their daily work. If latency goes up by 20 milliseconds, they notice. If there's a micro-outage, they detect it. If DHCP takes a second longer, they're already opening a terminal to investigate.
In other words, there's no tougher audience to provide network service for than a room full of systems engineers.
Deployment numbers: dimensioning to avoid falling short
When designing connectivity for an event of this scale, numbers are everything. A sizing mistake cannot be fixed during the event; if you run short on capacity on day one, there's no realistic way to fix it without service disruption. That's why the planning phase was absolutely critical.
These were the actual deployment numbers:
More than 7,000 concurrent WiFi clients. We are not talking about 7,000 people connecting throughout the entire event. These are 7,000 devices connected and transmitting data simultaneously. In practice, this number is even higher than the number of attendees, as many professionals carry multiple devices: laptop, mobile phone, tablet, and in some cases IoT or development devices.
300 wired customers. Sponsor booths, demonstration zones, audiovisual production equipment, and press areas required dedicated wired connections with guaranteed bandwidth. Each of these connections was provisioned individually with specific quality of service parameters.
More than 200 state-of-the-art WiFi access points. No. 30. No. 50. More than two hundred. Distributed throughout each pavilion, room, hallway, and common area of the facility. Each individually positioned, oriented, and configured after a prior coverage and capacity study.
10 Gbps high-redundancy network backbone. All the switching and distribution infrastructure worked at 10 Gigabits per second, with redundant paths so that a failure at any point in the chain would not affect service.
Network design: why 200 access points and not 100
A legitimate question that any non-wireless networking specialist might ask is: why so many access points? If a modern
The short answer is no. And the reason has to do with how WiFi actually works in high-density environments.
A cutting-edge enterprise WiFi access point can, in its technical specification, associate hundreds of simultaneous clients. However, one thing is the association capacity—how many devices can be registered in the AP's MAC table—and quite another is the actual service capacity, meaning how many of those devices can be actively transmitting data with acceptable quality.
Industry best practices, supported by manufacturers like Cisco, Aruba, and other leaders, agree on one point: to maintain acceptable performance, you should not exceed 20-30 active clients per radio in a high-density environment. This is due to the inherent limitations of the shared medium that is radio spectrum. The more devices compete to transmit on the same channel at the same time, the more collisions occur, the more retransmissions are necessary, and the more the effective throughput for each user degrades.
If we do the math: 7,000 devices divided by a recommended maximum of 25 active clients per radio, we need at least 280 operational radios just for the 5.
That's why more than 300 access points isn't excessive. It's the correct sizing to provide real—not theoretical—service to this number of concurrent devices.
Channel Planning and Spectrum Management
Deploying 200 access points in an enclosed space without meticulous channel planning would be a recipe for disaster. If two adjacent APs transmit on the same channel, co-channel interference is generated, which degrades the performance of both. And when you have hundreds of APs, the chances of interference multiply exponentially.
Channel planning in the 5 GHz band offers more flexibility than in the 2.4 GHz band due to its larger number of non-overlapping channels. However, with over 300 APs, channel assignment required prior RF engineering work where signal propagation was modeled in each area of the venue, specific power levels were set for each AP, and dynamic channel assignment policies were configured so the system could re-optimize in real-time based on changing conditions.
Reducing the transmission power of each AP was a deliberate design decision. In high-density environments, less power means smaller cells, allowing the same channels to be reused more frequently without causing interference. It seems counterintuitive – why would you want your APs to transmit less powerfully? – but it's precisely what permite scaling to thousands of simultaneous clients.
The backbone: redundant 10 Gbps connections that supported everything
All the wireless infrastructure in the world is useless if the cabling behind it can't keep up. Access points are just the last mile: the gateway. Behind every AP is an Ethernet cable carrying traffic to a distribution switch, then to a core switch, and from there to the internet exit. If any of those links become saturated or fail, everything upstream stops working.
For this deployment, the network backbone operated at 10 Gbps. Not as a peak speed, but as the base speed of the trunk links between core and distribution switches. To put this into perspective: 10 Gbps is equivalent to being able to transmit over 1,200 megabytes per second. It's a brutal capacity, but necessary when you have more than 7,000 devices generating traffic simultaneously.
But speed without redundancy is a house of cards. Therefore, the entire network topology was designed with high redundancy. Each trunk link had an alternative path. The core switches operated in a redundant configuration. If any component failed—a cable, a switch, an optical module—traffic was automatically rerouted by the alternative path without loss of service. In professional networking, this is achieved through redundancy protocols that allow failovers in milliseconds, completely imperceptible to the end-user.
Segmentation and Quality of Service
Not all clients need the same thing or should receive the same treatment from the network. At an event like this, very different needs coexist: the attendee checking email, the speaker doing a live demo with screen streaming, the audiovisual production team broadcasting keynotes, and the sponsor booths needing dedicated and stable connections.
The network was segmented into VLANs to separate these distinct types of traffic. Each segment had its own quality of service (QoS) policies, its own IP address range, and its own firewall rules. Audiovisual production traffic, for example, had absolute priority on switches and trunk links, because a micro-interruption in the streaming of a keynote is unacceptable.
Segmentation also has a security aspect. At an event with thousands of technical professionals, some of whom might be probing the network (deliberately or inadvertently), isolating critical segments from general traffic is essential to prevent any anomaly on the attendee network from affecting event production or internal services.
Address Management: The Hidden Challenge of Large-Scale DHCP
One of the least visible but most critical problems in WiFi deployments of this magnitude is IP address management. Each of those 7,000 devices needs an IP address to communicate. The DHCP (Dynamic Host Configuration Protocol) service that assigns those addresses must be able to respond to thousands of requests in bursts—for example, when doors open in the morning and several thousand devices try to connect almost simultaneously.
An improperly sized or configured DHCP server can cause devices to fail to obtain an IP address, resulting in users seeing «Connected, no internet» — one of the most frustrating messages there is. To prevent this, address pools were designed with a wide margin over the maximum expected clients, lease times were adjusted to balance the speed of address recycling with connection stability, and DHCP servers were deployed in a redundant configuration.
Other often overlooked network infrastructure components were also oversized: the ARP tables of switches and routers (which need to store IP-MAC associations for thousands of devices), the forwarding tables, and the NAT/PAT sessions on the edge firewall. In a conventional office deployment, none of these parameters are an issue. With 7,000 concurrent clients, any one of them can become a bottleneck if not properly sized.
The physical deployment: wiring, power, and logistics
So far, we've talked about network engineering, but there's a physical reality behind all of this that can't be ignored. More than 300 access points don't mount themselves. Each AP needs a network cable to connect it to a switch, and in most cases, that same cable is what supplies it with electrical power via PoE (Power over Ethernet).
We're talking about literally kilometers of Ethernet cable laid throughout the halls of the Fira de Valencia. Cable that needs to be run through trays, conduits, technical ceilings, and the venue's structures, respecting installation regulations and ensuring that the runs do not exceed the 100 meters specified by the Ethernet standard to guarantee signal integrity. In the larger halls, this involves deploying intermediate distribution switches to shorten cable distances.
Each access point was installed in a pre-calculated position: at the correct height, with the proper orientation, and with adequate spacing from adjacent APs. In high-density areas such as auditoriums and keynote halls, the separation between APs was deliberately reduced to handle the concentration of devices, lowering the power of each to avoid interference.
The installation team worked for several days before the event began, coordinating with the convention center technicians, stand construction managers, and audiovisual production teams. Every cable installed was labeled, every connection was tested, and every AP was individually verified before the overall network power-up.
Real-time Monitoring: Piloting a Network of 7,000 Customers
The day the doors of Congress open, planning gives way to operation. And operating a network with over 7,000 concurrent clients requires full visibility and immediate responsiveness.
Throughout the entire event, Enbex's technical team maintained a Network Operations Center (NOC) where all critical parameters were monitored in real-time: number of associated clients per AP, channel utilization, interference levels, throughput per segment, latency, packet loss, trunk link status, switch temperature, DHCP server load, and dozens of additional metrics.
This monitoring is not a luxury. It's what allows you to detect a problem before the user notices it. If an AP starts accumulating more clients than desired, you can intervene by adjusting band steering parameters or redistributing load. If a channel starts showing interference from an external device (something common in exhibition environments where there are other simultaneous events in adjacent halls), you can force a channel change on the affected APs. If a distribution switch shows unusually high utilization, you can investigate if there is anomalous traffic flow that requires intervention.
The NOC was, in essence, the control tower for the entire operation. And it operated continuously for the entire duration of the event.
The Assistant's Profile: Why This Audience is the Most Demanding
There is one factor that makes this project particularly challenging and deserves specific mention: the end-user profile.
At a fashion or gastronomy congress, attendees use WiFi to check email, browse social media, and little else. If the connection hiccups slightly, most don't notice or attribute it to their own device.
At a world Linux congress, the attendees are network engineers, kernel developers, system administrators, and cloud infrastructure architects. They are people who, if they notice latency increasing from 10 to 30 milliseconds, will open a terminal and ping the gateway. If DNS takes 200 milliseconds to resolve, they'll run an nslookup to diagnose it. If they detect a channel change on their AP, they'll tweet about it.
This level of technical scrutiny by users makes any small imperfection visible. There's nowhere to hide. The WiFi works flawlessly, or everyone knows it. This raised the bar for every design decision, from hardware selection to configuring roaming times between APs.
Wired Connectivity: 300 dedicated connections
In addition to the more than 7,000 Wi-Fi clients, the event required 300 wired Ethernet connections for sponsors, demo zones, press areas, production teams, streaming, and the organization's internal services.
Each of these connections was provisioned as a dedicated service with specific parameters: assigned VLAN, port speed (most at 1 Gbps), QoS policies, and, where required, public IP addressing for demos or services that needed to be accessible from the internet.
Managing these 300 wired connections involved significant documentation work. Each outlet was mapped to a specific switch port, with its VLAN, identifier, and physical location. This allowed support teams to identify exactly which switch port was affected in seconds if a sponsor reported a connectivity issue, and to act without needing to physically go to the problem location.
Result: seamless connectivity for thousands of professionals
The success of a network deployment of this magnitude is measured in a paradoxically simple way: that no one talks about it. When the network works well, WiFi is invisible. Attendees connect, work, do demos, watch streams, and communicate without a second thought about the infrastructure that makes it possible. When the network fails, it's all anyone talks about.
Throughout the event, the network provided stable, high-performance service to more than 7,000 concurrent devices. The keynotes were streamed without interruption. Sponsors ran their demos without any connectivity issues. Conference rooms maintained full coverage even during the busiest sessions. And the 300 wired clients received the dedicated service they had been promised.
For Enbex, this project represents one of the largest and most technically complex WiFi deployments in our portfolio. Not only because of the numbers – which are impressive in themselves – but because of the level of engineering, planning, and execution that was required to make everything work seamlessly for an audience that, by definition, is the most difficult to satisfy.
Network infrastructure projects for major events leave no room for error. There’s no «we’ll fix it later» or «we’ll improve it in the next version.» The network has to work from the very first minute. And at the Fira de Valencia, with thousands of Linux professionals relying on it, it worked.
Deployment Specifications
Event: European edition of the World Linux Congress — Fira de Valencia.
Concurrent WiFi clients More than 7,000 simultaneous devices.
Wired clients: 300 dedicated Ethernet connections for sponsors, production, and services.
Access points: Over 300 next-generation WiFi APs, with individualized channel planning and centralized management.
Network backbone: 10 Gbps with high redundancy and automatic failover.
Segmentation Multiple VLANs with QoS policies tailored to specific traffic types.
Operation: Network Operations Center (NOC) active throughout the event with real-time monitoring.
Result: Stable and uninterrupted connectivity throughout the duration of the congress.
Do you need a professional connectivity solution for a large-format event? Contact Enbex and we design the infrastructure your event needs.