OnyxPlay · Deployment & Integration

Six weeks from handshake
to first subscriber.

BGP-anycast CDN, on-network edge hardware, direct peering with your ASN. You keep routing control. We run the software stack.

Start the conversation See the appliance spec
6 weeks
MoU to first subscriber
standard partner timeline
BGP-anycast
Scale-out topology
N nodes · you keep routing control
40 Gbps
Throughput per node
~95% cache hit · 60 K subs
24×7
NOC coverage
Mumbai operations · English + Hindi
Commissioning timeline

The actual production path. Click any phase.

Edge appliance

OnyxPlay CDN — 1U bare-metal node.

Signed install image · OnyxPlay-operated · BGP-anycast across N nodes. You own the hardware; we operate the software stack.

60 K
subs / node
40 Gbps
throughput
24 TB
cache / node
N nodes
anycast
Reference compute
CPU
AMD EPYC 9354P · 32 core · 3.25 GHz (or equivalent)
Memory
256 GB DDR5 ECC · 4800 MT/s
Boot
2× 480 GB NVMe · mirrored
Reference cache
Tier 1
4× 3.84 TB NVMe Gen4 · ~95% hit
Tier 2
6× 7.68 TB SATA SSD · long-tail catalogue
Total
~61 TB raw · ~24 TB usable after parity
Network & BGP
Data NICs
2× 25 GbE SFP28 · LACP bond to your BGP switch
Routing
BGP4 speaker on every node · anycast member
OOB / BMC
1× 1 GbE · IPMI 2.0 + Redfish for our installer
Power & env
PSU
Dual 800 W 80-Plus Platinum · A / B feed
Draw
380 W typical · 540 W peak (per node)
Cooling
Front-to-back · 35 °C cold-aisle max
Site prerequisites

What you bring to the install. Six items.

01
Server hardware

N servers to OnyxPlay reference spec (1U each) · procured by you from any tier-1 OEM

02
Rack space

1U slot per node · 19″ rails · 760 mm depth · 35 °C cold-aisle max

03
Dual power

A + B feeds per node · 16 A IEC-C13 per PSU · UPS or generator-backed

04
BGP router

Each node patches into a router that accepts BGP4 anycast announces from the node

05
IP & ASN

/29 IPv4 + /112 IPv6 per node · your ASN for peering · reverse-DNS under your zone

06
Ops contacts

2× 24×7 NOC contacts · 1× escalation owner · IPMI / serial access for our installer

Connectivity options

Two paths. PNI preferred. IXP supported.

PNI Preferred
Private Network Interconnect

Cross-connect in our cage at any of our six PoPs. Dedicated capacity, no shared-fabric contention, deterministic latency.

Setup
5–8 business days
Capacity
10 / 25 / 100 GbE
Latency
~0.4 ms RTT in-metro
Best for
Steady, high-volume traffic · partners > 8 Gbps peak
IXP Supported
IXP peering

Bilateral or via route servers — both supported. We peer on DE-CIX, Extreme-IX, and NIXI; three IXPs × two cities = six on-ramps.

Setup
10–14 business days
Capacity
Shared fabric
Latency
0.6–1.2 ms RTT
Best for
Smaller partners · already on the IX
IXP presence: DE-CIX Mumbai · DE-CIX Delhi · Extreme-IX Mumbai · Extreme-IX Delhi · NIXI Mumbai · NIXI Delhi
Day-2 operations

NOC-watched. 24×7. Mumbai.

Eight live metrics, auto-paged on threshold breach. The partner portal shows your node's dashboard in real time.

node-cluster · live all systems nominal
CACHE_HIT
96.4%
rolling 1h · target ≥ 90%
EDGE_RTT_P95
1.8 ms
subscriber to nearest node
STREAM_FAILS
0.12%
5xx + DRM failures · target ≤ 0.5%
CONCURRENT
28.4 K
71% of provisioned capacity
BGP_SESSIONS
3 / 3
anycast members up · 6 d 14 h
ANYCAST_DRIFT
< 1 %
cross-node session leak
CPU_LOAD
42 %
avg across 32 cores · per node
PSU_REDUNDANCY
A + B
all feeds present · 3 / 3 nodes
Recent alerts
14:02:11 Cache pre-warm completed for Cricket Live HD across 3 nodes (T20 final)
13:48:03 Rolling update v2026.04.18 applied node-by-node · 0 stream interruptions
11:21:54 node-02 BGP flap (4 s) · anycast peers absorbed traffic · no subscriber impact
09:00:00 Daily MIB Schedule 10 returns submitted on your behalf
07:14:22 Widevine license keys rotated across nodes · 0 client renewals failed
SLA matrix

Response + resolution times. In writing.

99.95%
Anycast pool availability
monthly rolling · per region
99.99%
Origin availability
monthly rolling · headend
< 250 ms
Channel-change time
p95 · LL-HLS
24×7
NOC coverage
Mumbai operations
TierDescriptionResponseResolveCredits
P1 · Service down Anycast pool unreachable, > 5% concurrent subs affected. 15 min 2 h 5% of monthly fee per 30 min over SLA
P2 · Degraded Single node down with anycast holding, channel failing. 1 h 8 h Case-by-case
P3 · Functional EPG mismatch, billing webhook lag, single-subscriber issue. 4 h 2 business days None
P4 · Cosmetic Portal label, app screenshot refresh, dashboard tweak. 1 biz day Next release None
Deployment FAQ

Questions before the deployment call.

Answered by our deployment team on every intake call.

Get the deployment checklist
No — you procure servers to our published reference spec from any tier-1 OEM (Dell, HPE, Supermicro, Lenovo are all certified). We ship a signed install image and drive the install remotely over your OOB.
One node serves up to 60 K subscribers at ~40 Gbps. Above that, or any time resilience matters, you run a 3-node anycast cluster — BGP handles failover automatically.
Every node announces the same anycast prefix into your network with a distinct BGP community per node. You adjust your local-pref, MED, or community-based policy to steer specific subscriber pools to specific nodes.
A signed peering agreement under our standard template, an IRR-registered prefix list for your AS, and a maintained PeeringDB entry. PNI is preferred; IXP peering is the fallback.
Our 24×7 NOC owns Tier-2 and Tier-3 from the moment the first node passes burn-in. You own Tier-1 (subscriber-facing support) with scripts we provide. We sit on a war-room bridge for the first 48 hours after public launch.

Ready to rack the first node?
Let's schedule the kickoff.

We'll send you the reference hardware spec, the IP/ASN intake form, and the deployment playbook in the same day.

Start the conversation Compliance details