Sizing Collabora Online servers
Working out CPU threads, memory and bandwidth for Collabora Online from total and concurrent users with the vendor's per-user figures, plus headroom for large spreadsheets and presentations.
Collabora Online scaling and sizing
Sizing Collabora Online servers for real concurrency and scaling beyond one server with WOPISrc-aware load balancing and the COOL Controller on Kubernetes.
Scaling Collabora Online on Kubernetes with the COOL Controller
Using the COOL Controller for memory-based scaling and document migration between pods: the umbrella chart, leader election, scale-down at night and weekends, and when a fixed replica count is enough.
Load balancing Collabora Online with WOPISrc routing
Configuring HAProxy or Nginx so every request for a document lands on the same Collabora server: url_param WOPISrc balancing, tunnel timeouts for WebSockets, and what breaks without it.
Running multiple Collabora Online servers without Kubernetes
Scaling Collabora Online across several VMs or bare-metal servers behind a document-aware load balancer: routing, health checks, rolling restarts, and capacity planning per server.
Collabora Online capacity for large collaborative sessions
What limits many people editing one document at once in Collabora Online, how meeting-style sessions of around 20 editors behave, and settings and sizing for town halls, classrooms and workshops.
