The shortest path is a design decision
Delivery performance is decided by the network path long before it is decided by the cache. Peering, placement and transit shape the experience more than any amount of application tuning can.
The network is ours, not a reseller account
PLAYBOOX is a streaming platform powered by Cloud Point, an infrastructure operator with its own autonomous system, its own address space and its own presence in Dhaka. Most platform vendors rent all three from someone who rents them from someone else. Here is the thing that follows from owning them: when delivery needs to change, we change it, rather than filing a ticket with whoever actually controls the path.
Our own
Held directly
Panthapath
Multi-homed
1 Tbps
RPKI valid
BDIX peered
- Cloud Point holds its own AS number and announces its own routes. Nothing in the delivery path is subleased from a hosting provider who could change it without asking.
- Registered to the company at APNIC rather than borrowed from an upstream. Addresses we assign are addresses we control.
- Dhaka-1205, alongside the BDIX core exchange. Being in the building where the peering happens is what makes in-country delivery short rather than nominally local.
- More than one transit network carries the traffic, so a single provider going down is not the same thing as the platform going down.
- One terabit per second across the backbone. Against the 2.885 Mbps this platform actually measures as its delivered average, that is headroom for roughly 346,000 concurrent streams before the backbone becomes the constraint — which is the useful way to read a capacity number, because on the night it matters the question is never how big the pipe is, it is what runs out first.
- The routes are RPKI-signed and IRR registered, which is what stops somebody else announcing our address space. Most networks this size have not bothered.
- A viewer in Bangladesh is served over local peering instead of an international transit link, which is most of the difference between a stream that starts instantly and one that does not.
The streaming platform is the top of the stack, not a tenant on someone else’s
Streaming platform
Platform layer
Cloud infrastructure
Network
What the parent operates
None of this is resold. It is the same estate the streaming platform runs on, which is why a deployment can be sized against real hardware rather than against a quota.
Compute
Virtual machines and Xeon-class dedicated servers with ECC memory, NVMe storage and dedicated uplink.
Object storage
S3-compatible and encrypted, which is what lets a VOD origin sit in-country instead of in a public cloud bucket.
Managed Kubernetes
Cluster management, worker pools and load balancing, for the application tier that does need to scale out.
Managed operations
Provisioning, monitoring, security hardening and incident support, on the same estate rather than through a vendor queue.
Business email
Custom domains with DKIM and SPF, run alongside everything else rather than bought from a third party.
Connectivity
BDIX and Coloasia, from a facility in Panthapath that sits beside the exchange rather than a few hops from it.
Staffed operations
The facility is attended around the clock. That is a statement about people on site, not a promise about uptime, and the two get confused often enough to be worth separating.
Verifiable, just not in public
Panthapath, Dhaka-1205 · info@cloudpoint.com.bd
One frame, end to end
Two paths in, one protected path out. This is the route every second of video takes — from a camera or a master file, through our own encoding and our own delivery network, to a screen. Nothing in the chain is rented.
Live
A broadcast feed arrives over standard contribution protocols, is handed off with minimal delay and repackaged into encrypted segments as it runs.
On demand
A master file is transcoded once into a full quality ladder, encrypted, and published to origin storage ready to be served forever after.
The return path
Every player reports what it actually experienced — start-up time, rebuffering, errors. That data comes back to us continuously and informs how the ladder is tuned.
Peered inside Bangladesh, across every BDIX zone
PLAYBOOX connects straight into BDIX — the country's largest internet exchange — across seven zones nationwide. A viewer's stream reaches them over local peering instead of crossing an international link, so it arrives in milliseconds and plays without buffering. Traffic that starts in Bangladesh stays in Bangladesh.
One hop away
Each city is served from its nearest BDIX zone, so a stream takes the short path — not a detour through another continent.
No international transit
Local peering means playback does not ride a costly overseas link or anyone else's congestion. Fewer hops, fewer things that can go wrong.
Built for the spike
Capacity follows demand automatically and traffic reroutes across zones, so a sudden surge — a match, a premiere — is absorbed instead of buffered.
Buffer-free by design
Short local paths, edge caching and an adaptive bitrate ladder keep the picture smooth from the first frame through the last.
A cloud we can walk up to
The entire platform runs on hardware we own and physically hold. Not a rented region, not a shared tenancy — a hypervisor on our own rack, carved into isolated machines and declared, every line of it, in version control.
Each service gets its own virtual machine or container: its own kernel, its own network address, its own failure domain. A misbehaving encoder cannot reach the billing database, and a rewrite of one service never means redeploying another.
Transcoding runs GPU-accelerated on the same node, so a master file never leaves the building to be encoded. Origin storage is S3-compatible, which means the day it makes sense to move it to a public cloud, the services reading from it do not change a line.
- Bare-metal, self-hostedno rented instances anywhere in the path
- 96 vCPU · 156 GB RAMa single 2U node runs the whole estate
- 18 separate guestsone service, one failure domain
- GPU passthroughhardware transcoding on the same box
- S3-compatible originportable to any object store
- 22 Terraform modules≈5,600 lines, all in version control
Three ways it can run
On our metal
The platform as it runs today — our hardware, our network, our operational responsibility. Lowest cost per stream and nothing to negotiate with a vendor.
On yours, on premises
The same stack, declared in the same Terraform, stood up inside your own datacentre. Useful where content, data residency or regulation cannot leave the building.
Lifted to public cloud
Nothing in the stack is tied to our hypervisor. Object storage is S3-compatible and every service is containerised, so a migration is a change of target, not a rewrite.
Where does your video traffic exit today?
Transit volume, peering arrangements and where your audience actually sits. Those three answers usually reveal the cheapest available improvement.