---
title: "Filmin: 256 vCPU instances for a streaming platform that doubled its subscribers"
description: "How Filmin, a Barcelona streaming platform, runs on 256 vCPU Noraina instances in Paris behind Cloudflare, with disaster recovery in Dublin."
url: https://www.noraina.cloud/customers/filmin/
language: en
---
# Filmin: 256 vCPU instances for a streaming platform that doubled its subscribers

Filmin is a subscription streaming platform based in Barcelona, serving Spain, Portugal and Mexico. When its subscribers doubled and hours streamed grew tenfold in 2021, we moved its applications to 256 vCPU instances on our platform. They run in Equinix PA2 and Telehouse TH3 in Paris, behind Cloudflare, with disaster recovery in Equinix DB3 in Dublin.

## The problem

In 2021 Filmin grew fast. New content and lockdowns that kept people at home doubled its subscribers in twelve months, and hours streamed grew tenfold. The load exposed the limits of the infrastructure underneath. Filmin wanted better performance and a lower marginal cost for every hour of video it delivers.

At the time, the platform handled around 100,000 requests a minute from 2.5 million unique IP addresses a month, and Cloudflare blocked 32 million threats a month.

Filmin's applications are tightly coupled to their datastores, so the latency between application server and datastore decides how the service performs. Growth needed both scale-up and scale-out: large instances, and a fast network to carry the replication between them.

## How the pieces fit

1. **The instances.** After benchmarks and load tests, the best performance came from running each datastore on the same instance as its application, and replicating content between instances to scale out. We provided instances with 256 vCPUs, 512 GB of RAM and replicated solid-state storage, on a redundant 25 Gb/s network with 50 Gb/s available per server. Production runs in two data centres in Paris, Equinix PA2 and Telehouse TH3, with the connectivity and low latency Filmin's subscribers need.
2. **Application security.** As a Cloudflare partner, we put Cloudflare in front of the platform to filter bots, DDoS attacks and other malicious traffic before it reaches the servers.
3. **Disaster recovery.** A disaster recovery environment runs in Equinix DB3 in Dublin, in a different country from production, so the service can be brought back if the Paris sites fail.

## What you get

- Services that stay stable and responsive during peaks of use.
- Fewer, larger servers: Filmin's previous bare-metal provider topped out at 72 vCPUs per server, and for a little more Noraina provides a managed 256 vCPU server. With fewer datastores to run, management costs go down.
- Bot and DDoS protection from Cloudflare, managed by our team.
- Production in two Paris data centres and a disaster recovery environment in Dublin.

## Cost

When Filmin compared a three-year reservation and the monthly running costs in 2021, the Noraina solution cost less than half of the same service at one of the major cloud providers. Instance prices are on the [Efficient Cloud Instances](https://www.noraina.cloud/products/efficient-cloud-instances/#pricing) page.

## Frequently asked questions

### Which infrastructure does Filmin run on with Noraina?

Filmin runs on Noraina instances with 256 vCPUs, 512 GB of RAM and replicated solid-state storage, on a redundant 25 Gb/s network with 50 Gb/s available per server. Production runs in Equinix PA2 and Telehouse TH3 in Paris, the disaster recovery environment is in Equinix DB3 in Dublin, and the platform sits behind Cloudflare application security.

### Why did Filmin need instances with 256 vCPUs?

Filmin's applications are tightly coupled to their datastores, so latency between them limits performance. Our benchmarks and load tests showed the best results with the application and its datastore on the same large instance, replicated to scale out. Filmin's previous bare-metal provider offered at most 72 vCPUs per server.

Last updated: 2026-10-12