The infrastructure layer for what comes next.
Joy Services builds and runs the layer the internet stands on: fiber we own, data centers we are building, racks, routing, storage, edge delivery, intelligence and devices — on one platform with one console and one API, serving more than ten thousand enterprise customers around the world. Not a reseller. Not a dashboard on somebody else's cloud. The real thing, built from the metal up and designed to your demand.
One cloud, every layer, owned end to end.
Most "clouds" are a control panel over somebody else's data center. We are building the whole stack: the fiber in the ground, the data centers, the racks, the switches, the addresses, the hypervisors, the storage fabric, the DNS, the edge, the devices and the software that runs them — so that every layer answers to the same design and the same people, and every layer can be shaped to what a customer actually needs.
Physical
Fiber we own in Delhi, Mumbai and Chennai, racks in facilities we operate in, two Joy data centers on the way, interconnection inside the data center and across the world. The ground the rest stands on.
Compute
Virtual machines, dedicated servers and colocation on hardware in our own racks. Windows and Linux, from a single vCPU to a full cabinet, provisioned by the same automation — and custom infrastructure designed to a customer's demand when a menu is not enough.
Network
Our own address space, our own routing, our own switching fabric, our own exchange presence. Traffic stays on infrastructure we control from the customer's port to the internet exchange, in every region, and connects to the networks our customers already run.
Storage
Distributed, self-healing storage that survives disk, node and rack failure without a human in the loop. Block for machines, object for applications, snapshots for everyone.
Security & observability
Isolation by default, audit on everything, identity that follows a customer across every service, and real-time monitoring of every host, port and flow — so trust is a property of the platform rather than a feature you buy.
Intelligence & devices
Inference where the data is — on the customer's own machines and devices at the edge — with our cloud holding the shared memory; brokers, private APNs and telemetry for the sensors, meters and machines of the Internet of Things.
One console, one API
Every service above is ordered, billed, monitored and automated from a single console and a single API, for direct customers and for partners who resell under their own brand.
One network, edge to core, on every continent.
Our own autonomous system on our own address space, carried over fiber we own between our metros, lit at exchanges and landing stations, and interconnected inside the data center and across the world. Turn the globe: filled points are lit today, hollow ones are planned, green lines are fiber we own.
What actually happens when you click.
Every button in the console is a pipeline we wrote and run. These are the ones that matter most, drawn the way they work: deploy, protect, store, deliver, route and watch.
Order to running machine in about ninety seconds
A verified template is cloned, the disk grown, a static address assigned with reverse DNS, the machine booted and the platform agent inside it sets the credentials — then the console and the API light up.
- 01
Place
The host with free capacity is chosen; the disk is spread across the storage pool.
- 02
Clone and configure
Template copy, disk, IP, hostname — no hand-built images, nothing e-mailed.
- 03
Boot and hand over
The agent confirms the network is up; power, reinstall, snapshot, resize and console are live from that second.
Attacks are diverted, scrubbed and forgotten
Every flow at every border is scored in real time. When a score crosses the line, the affected prefix is diverted to the scrubbing centre automatically — Equal Diversion keeps the address reachable while the attack is dropped — and legitimate traffic lands as before.
- 01
Score
Source, destination, ports and volume on every border router, continuously.
- 02
Divert
BGP diversion in seconds, no ticket and no phone call; GRE delivery for prefixes hosted elsewhere.
- 03
Deliver
Real users reach the same address; a blackhole community or an Urgent ticket for the extreme case.
A disk dies and nobody notices
Every write is acknowledged only once it exists on several nodes in more than one rack. When a disk, a node or a rack fails, the missing copies are rebuilt on healthy hardware by the system itself.
- 01
Write once
Through the S3 API or a volume attached to your server.
- 02
Copies in two racks
Independent power and switching; no rack holds all copies.
- 03
Rebuild
Automatic, from the surviving replicas, without a maintenance window.
One address, every edge, one request to your origin
The same address is announced from every edge, so a visitor lands on the nearest one. Misses collapse into a single shield path, a purge clears every edge in seconds, and the visitor's real IP reaches your logs.
- 01
Anycast
The nearest edge answers; static content never touches the origin.
- 02
Shield
Many edges, one upstream request.
- 03
Purge
URL, prefix or everything — gone from every edge in seconds.
Your prefixes, our network, deterministic paths
Sessions with AS152565 over a cross-connect, an exchange port or a GRE tunnel; prefix filters built from your IRR AS-SET and refreshed nightly; RPKI-invalid routes dropped at every border; billed at the 95th percentile.
- 01
Bring an ASN
Or take addresses from Joy space — LOA and RPKI ROAs for your own.
- 02
Peer or buy transit
Bilateral sessions above 1 Gbps, route servers otherwise — the peering portal has the guide.
- 03
Stay protected
DDoS-protected transit on your prefixes; tag an address to blackhole it anywhere.
Every host, every port, every 15 seconds
A collector in every data center reads the hypervisors and switches on its own floor and reports to the core. The console shows your servers' live figures; the status page flips a region to degraded on its own when a host stops answering — and says so before the ticket queue does.
- 01
Collect
CPU, memory, disk, network, ports and flows, locally in each region.
- 02
Decide
Components marked "auto" change state from the telemetry, not from a human.
- 03
Tell
Your server page, the public status page and /status.json, within the same minute.
Every request takes the same path, everywhere.
The platform is built in tiers that can each be multiplied on their own. Add a frontend when a region gets busy, add a cache node when a data center comes online, add hardware when customers need it. Nothing is a single box, and nothing is special.
Frontends
Many small web nodes close to the customer. They render the site and the console and hold no state of their own.
- Any number, any region
- Stateless, replaceable in minutes
- Real visitor IP preserved
API & cache
The one place logic lives. Every read is served from memory first; every write is a job with a log line.
- Memory-first reads
- Job queue for everything slow
- One API for console, partners, staff
Regional cache nodes
One per data center, talking to the hypervisors and switches on its own floor and reporting back to the core.
- Local view of every host
- Survives a link to the core going down
- Bootstrapped from the console
Hypervisors, switches, storage
Our own hardware in our own racks: compute hosts, the switching fabric, the distributed storage cluster.
- Placement spread across storage
- Templates verified per host
- Self-healing storage
Database & ledger
Written only through the core, never directly. Every action, price, payment and change is a row somebody can audit.
- Single writer path
- Full audit ledger
- Backed up off-site
Why it matters to you: a busy region gets more frontends, not a slower site; a new data center is a cache node plus hardware, not a new platform; and every number in your invoice traces back to a logged action.
The next decade of software will be decided by who owns the layer underneath it.
Every company is becoming a software company, and every piece of software runs on somebody's infrastructure. When that infrastructure is rented, leased and abstracted three times over, you inherit every one of those layers' prices, outages and decisions. We think the world needs more operators who own their stack — the fiber, the building, the rack, the route — run it in the open, price it like a utility and shape it to the customer in front of them. That is the company we are building.
- 01
Own the metal
Fiber, servers, switches, storage and address space that belong to us. Nothing critical rented from a hyperscaler.
- 02
Build, don't wrap
Our control plane, our automation, our billing, our DNS. Third-party software only where it is best in class, and always behind our own API.
- 03
Long horizon
Decisions made for the next ten years, not the next quarter. Infrastructure is a trust business, and trust compounds slowly.
- 04
Honest by design
Real status, real audit trails, real confidence scores. If something is wrong, the console says so before a customer has to ask.
- 05
Global from day one
Designed for many regions, many currencies and many partners, so the same platform serves a shop in one city and an enterprise on four continents.
- 06
Shaped to the customer
Custom infrastructure, custom services and solutions designed to a customer's demand are the normal case, not the exception.
Tell us the problem. We build the layer for it.
The catalogue is where most customers start. Many finish somewhere the catalogue does not list: infrastructure designed around their workload, networks that reach into theirs, devices that report to them, and people on our side who watch it with them. All of it on the same platform, the same console and the same invoice.
Designed to your demand
Private clusters, unusual shapes, specific hardware, your own addresses, your own racks — specified with you, built by us, run as one service.
Reaches into your network
VLANs, GRE, private interconnect, BGP sessions, BYOIP and cross-connects, so our platform is an extension of yours rather than an island.
Inside the data center and across the world
Cross-connects into the Joy fabric, exchange ports, and international capacity between our metros, Europe, Southeast Asia and North America.
One network, deployed where you are
Anycast addresses, edges and points of presence on our own autonomous system, so a service built here answers from the nearest place.
Watched every 15 seconds
Hosts, ports, flows and sessions, on your server page and on the public status page, with people who react to it around the clock.
Sensors, meters, machines, cameras
Brokers, private APNs, telemetry and edge inference for devices that must never sit on the public internet.
Built for one customer, run like a product
A managed stack, a migration, a reseller platform, a storefront on our API — scoped, delivered and operated by the people who run the platform.
One contract, one engineer who knows your name
Dedicated hardware, private networks, managed operations and an audit trail your team can read line by line.
From a cloud we run to a fabric the world builds on.
This is direction, not a launch calendar. Each phase is a capability we intend to own outright before we move to the next one. Order matters more than dates.
Own the core
A platform we control end to end: compute on our own hypervisors, automation from order to running machine, authoritative DNS, distributed storage, a console and an API that every later service plugs into.
- Compute, dedicated and colocation from one console
- Own DNS, storage and edge under one control plane
- Partner and reseller API
Own the path
Fiber we own between our metros, our routing, our exchange presence, our protection, our edge — sold as cleanly as a virtual machine.
- BGP, IX fabric and VLAN as self-service
- DDoS protection and edge caching in every region
- Private interconnect between customer sites
Own the floor and the edge
Joy DC Mumbai and Joy DC Noida come online on our fiber; devices and inference move to the edge, where the data is, with our cloud holding the shared memory.
- Two Joy data centers of our own
- IoT brokers, private APNs, telemetry
- Edge inference as a platform service
Own the map
Points of presence on every continent, joined by our network, running the same software, so a customer anywhere sees one platform with the same prices, the same API and the same guarantees.
- Regions on every continent
- One identity, one wallet, one console worldwide
- A marketplace where partners sell on our fabric
Slow where it must be. Fast everywhere else.
Infrastructure is the one part of technology where being boring is a virtue. Nobody wants an exciting power grid. So we take our time on the layers that have to last: the fiber route, the building, the network design, the storage fabric, the identity model, the audit trail. Those are decided once, for a decade.
Everything on top of them moves quickly. New services are modules, not new companies. A partner can resell what we build the day it ships. A customer can reach any of it from one console, one API and one wallet, in their own currency, wherever they are — and when the catalogue does not fit, we design what does.
We publish what is live, what is being built and what is still research, and we keep those three lists honest. When something breaks, the status page says so before the ticket queue does. That is the whole philosophy: own the layer underneath, run it in the open, and let people build on it without having to trust anyone but us.
Joy Services · Fiber, data centers, cloud, network, storage, security, intelligence and devices, on hardware we own
From a single machine to a whole business built on us.
The same console serves someone renting one virtual machine and a partner running a hosting brand of their own on our fabric. Nobody gets a different, lesser platform.
Ship on real hardware, pay like a utility
A machine in minutes, an API for everything the console does, clear prices in your own currency.
- Custom specs, not a menu of fixed sizes
- API tokens with scoped permissions
- Snapshots, reinstalls and console access
Sell everything we build, under your name
White-label the whole catalogue: your prices, your brand, your customers, our infrastructure and automation.
- Reseller API and wallet
- Marketplace listings for what you build
- Sub-partners under partners
Own the stack without running the data center
Dedicated servers, colocation, private networks, custom infrastructure and a control plane your team can audit line by line.
- Dedicated hardware and rack space
- Private VLANs between your systems
- Tax invoices wherever you are billed
Bring your addresses, peer on our fabric
Announce your own space, take an exchange port, protect it, and manage it from the same console as everything else.
- BGP sessions and IX ports
- DDoS visibility per block
- Reverse DNS on every address
Sensors, meters, machines and cameras that report to you
Brokers, private APNs, telemetry and inference at the edge for fleets of devices that must never sit on the public internet.
- Private cellular access into our network
- Per-device credentials and topic rules
- Learning shared across a fleet
A fabric to build products on
If your product needs compute, network, storage and identity in one API, on hardware nobody can pull out from under you, this is the layer.
- One API across every service
- Webhooks and events
- Marketplace distribution
Not numbers. Promises.
Uptime figures and latency charts are things you check on the status page. These are the things we will not compromise on, whatever the chart says this month.
- 01
No lock-in, ever
Your data, your images, your DNS zones and your backups can leave whenever you want, in standard formats. We keep customers by being good, not by making leaving hard.
- 02
The status page tells the truth
Incidents are posted when they start, not when they are fixed. If a region is degraded, the console says so on the page you are looking at.
- 03
Prices you can predict
Priced by spec, shown before you order, invoiced only after payment, in your currency. No surprise line items and no "contact sales" for a number.
- 04
Every action is logged
What was created, changed, billed or deleted, by whom and when, is a row you can read. Staff actions are in the same ledger as yours.
- 05
Tenants never see each other
Isolation is enforced at the hypervisor, the network and the storage layer, and verified, not assumed.
- 06
Partners get the same platform
What a reseller sells is exactly what we sell, from the same API, with the same guarantees. No "partner edition".
- 07
Built to your demand
When the catalogue does not fit, we design the infrastructure or the service that does, and run it with the same guarantees as everything else.
- 08
Support is run by operators
The people answering tickets are the people running the platform, around the clock. A ticket path exists for every problem, and it is answered by e-mail too.
Built, not bought
Our console, our automation, our billing, our DNS. When we do use outside software, it sits behind our own API and can be replaced without customers noticing.
Partners first
Everything we sell, a partner can sell under their own brand with their own prices, through the same API we use ourselves.
Priced like a utility
Clear prices, real invoices, local currency where it is wanted, and a wallet that works the same in every region.
Support that answers
A help desk run by the people who run the platform, with a ticket path for every problem and a status page that tells the truth.
The things we don't know yet, written down.
A vision is only honest if it includes the parts that are still hard. These are the questions the research list is organised around. When one gets answered, it moves to "building".
Can a new region be live in a day, from empty rack to first customer machine?
Bootstrap the cache node, register hardware, verify templates, join the network, appear in the console. Every step exists; making them one button is the work.
Can DDoS protection be a checkbox on an address block, priced like bandwidth?
Flow-level detection on our own routers, mitigation that starts without a phone call, and a bill that makes sense to a small customer.
Can object storage be as boring as disk, on hardware we own?
S3-compatible, self-healing across nodes and racks, and cheap enough to be the default place a backup or a snapshot goes.
Can a fleet of devices learn together without any of them sending raw data?
Inference on the device, memory and training in our cloud, learning shared across the fleet — with the privacy line unchanged.
Can a data center of our own be lit as one more node of the platform, the day it opens?
Fiber in, cache node up, hardware registered, templates verified, regions appear. Mumbai and Noida are where we find out.
Can a partner's whole business be one API key away from ours?
Brand, prices, customers, wallet, support, marketplace. A hosting company should be startable in an afternoon on top of this platform.
Every service. One platform.
5 live, 4 in beta, 5 coming — and the list keeps growing. Live services are in production. Beta services are provided as-is at the stated price. "Soon" services cannot be ordered yet; the Updates feed announces each one.
Cloud VPS
NVMe servers in every region, deployed in about 90 seconds.
LIVEBare Metal
Single-tenant Xeon / EPYC servers in our own racks, delivered in days.
LIVEStorage
S3-compatible object storage and NVMe volumes on hardware we own.
SOONCDN
Anycast edge delivery with origin shield and instant purge.
LIVEHosting
Managed PHP, WordPress and Node.js hosting with 1-click setup.
BETADatabase
Managed MySQL with failover and point-in-time recovery.
SOONAI
Inference at the edge, GPU compute in the rack, an operations assistant next.
SOONIoT
MQTT brokers, private APNs and telemetry ingestion.
SOONProxy
Private and dedicated IPv4 proxies on Joy address space.
LIVEDDoS Protection
Always-on scrubbing, BGP diversion and GRE delivery — included on every port.
BETAInfrastructure
Dedicated servers, colocation, IP transit and IX ports under one contract.
BETAColocation
Your hardware in our racks: per U or full rack, remote hands, cross-connects.
BETAAuthenticated SMTP relays and inbound processing for your domains.
SOONDNS
Authoritative DNS on our own name servers: every record type, transfers, API.
LIVENot on the list? Tell us what you need — custom infrastructure and custom services are built on the same platform.
Short answers to the obvious questions.
Anything not covered here has a person behind support@joycloud.services.
Are you a reseller of somebody else's cloud?
No. Compute runs on hypervisors in our own racks, on our own hardware, behind our own network and address space, over fiber we own between our metros. The console, the automation, the billing and the DNS are ours. Where we use outside software, it sits behind our API and can be swapped without customers noticing.
Where are you?
Everywhere our customers are. Compute regions in Noida, Mumbai, Chennai, Singapore and the US; points of presence and exchange presence across Asia and Europe; fiber we own in Delhi, Mumbai and Chennai; two Joy data centers under construction; more edges planned on every continent. Prices are shown in your currency with USD as the reference; tax invoices follow where you are billed.
What can I buy today?
Every service marked live in the catalogue above — virtual machines, bare metal, colocation, DNS, CDN, proxies and more — plus anything we design for you: custom infrastructure, custom services, interconnection, monitoring. The "Every service" list above is the catalogue itself, so it is always current.
Can I resell what you build?
Yes, and it is a first-class part of the platform, not an afterthought. Partners get the whole catalogue under their own brand and prices, through the same API we use ourselves, with a wallet, sub-partners and marketplace listings.
How do I peer with you, or bring my own network?
The peering portal has the whole guide: exchanges, direct sessions, filters, what we expect. Transit, IX ports, VLANs, GRE and BYOIP are on the Infrastructure page.
How do I get access, or talk to someone?
Open the console and register with an e-mail address. For partnerships, hardware, colocation, custom infrastructure or anything that needs a conversation, write to support@joycloud.services; it reaches the people running the platform.
What is the company behind this?
JOY SERVICES HOSTING PRIVATE LIMITED operates the cloud platform at joy.services, with JOY DATA SERVICES on the network side and JOY SERVICES IOT on devices and intelligence.
Build the layer underneath. Own it. Run it for decades.