When 1,000 people in a room all have their phones open, refreshing the same page, most platforms fall over. Pages load slowly, interactions feel sluggish, and the real-time excitement evaporates into frustrated buffering. EventWrist handles this differently — the architecture is designed from the ground up for sub-2-second synchronization across every device in the venue.
Real-time event engagement has a unique technical challenge: you need to sync hundreds or thousands of devices simultaneously, all receiving updates at roughly the same time, while also processing their inputs — votes, messages, check-ins. If every device polls a central server constantly, you get a thundering herd problem: the server gets overwhelmed at the exact moment when everyone is trying to participate at once.
Most event platforms solve this with a simple approach: they poll the server every few seconds and cache results. This works for small groups but breaks down as scale increases. The result is lag, missed votes, and messages that appear on some screens but not others.
EventWrist runs on Cloudflare Workers, Cloudflare D1, and Cloudflare R2. This stack gives us a globally distributed edge network rather than a single centralized server. When a guest sends a message or casts a vote, that data goes to the nearest Cloudflare edge node, which processes it and propagates it across the network in milliseconds.
Cloudflare D1 is a distributed SQLite database that replicates across Cloudflare’s network. Each event’s data lives in D1, and reads are distributed across edge nodes. When a guest checks in, their wristband code is validated against D1. When a vote is cast, it is written to D1 and immediately readable by the live screen queries.
D1 handles the relational data — wristband codes, event configurations, interaction settings, raffle eligibility rules. It is not designed for high-frequency real-time writes, which is why the system is architected to minimize writes during peak participation moments.
The critical insight for real-time sync is that the big screen does not need to read from D1 on every poll. Instead, the live screen state is cached at the edge and updated only when a meaningful change occurs. When a new vote comes in, the edge node updates the poll result cache. When a new danmaku message arrives, it is appended to the danmaku stream cache. These caches are invalidated and refreshed in under 2 seconds across all connected devices.
This approach means that even during peak moments — when 500 guests are all voting simultaneously — the system does not thrash. The caches absorb the burst, and the D1 database absorbs writes at a sustainable rate.
For the highest-frequency data — live check-in status, active danmaku messages, poll response counts — EventWrist uses Cloudflare KV. KV is a key-value store optimized for read-heavy workloads with eventual consistency. It handles the real-time stream of guest interactions without forcing every read to hit the database.
When a guest checks in, the check-in status is written to KV immediately. The Attendees dashboard reads from KV, not D1, which means it reflects the current state within seconds rather than having to wait for a database query.
Because the system is distributed across Cloudflare’s edge network, there is no single point of failure. If one edge node has issues, traffic is routed to the nearest healthy node automatically. The Cloudflare network spans 300+ cities worldwide, so even venues in remote locations get low-latency edge coverage.
For event organizers, this means you do not need to worry about infrastructure during your event. You do not need to provision extra server capacity for peak moments, because the edge scales automatically. You do not need to configure load balancers or optimize database queries, because that is all handled at the infrastructure level.
The guest experience is designed around a simple requirement: any modern smartphone with a browser. Guests do not need to install an app, grant permissions, or create an account. They scan the QR code, the wristband registers their device, and their browser communicates with the Cloudflare edge network. Everything happens in the browser — no native code, no app store, no installation friction.
The browser-based approach means the system works on iOS, Android, and any device with a modern browser. There are no compatibility issues with specific OS versions or device models.
Check-in is one of the highest-stress moments for any event platform — every guest arrives within a compressed time window and tries to check in simultaneously. EventWrist handles this by pre-generating all wristband codes before the event. Each wristband code is a unique identifier that maps directly to an event in D1. When the wristband is scanned, the check-in write goes to KV immediately, and the Attendees dashboard reflects the new state within seconds.
There is no authentication handshake, no OAuth flow, no session establishment — just a direct code scan that registers the device. This frictionless approach means check-in can happen in under 10 seconds per guest, even at scale.
The architecture is invisible to both hosts and guests, but it is what makes the experience feel effortless. To see this architecture in action from the host perspective, set up your first event and see how smoothly everything runs when 1,000 guests are all participating at once.
For a broader view of how EventWrist fits into your event technology stack, read the wristband check-in explainer.