XcellCloud|Edge Cloud
Compute where your data is made — in the store, the plant, the branch or the nearest metro.
XcellCloud Edge runs managed VMs and containers on edge nodes at your sites or in partner data centres* near your users, tied back to XcellHost's Mumbai cloud. Apps answer in milliseconds, keep working when the link drops, and send only what matters upstream.
The cloud, moved closer — without losing the cloud.
Edge cloud means running part of your application on small, managed compute nodes close to where people and machines are — a rack in your plant, a box in the back office of a store, or a server in a metro data centre — instead of sending every request to a central region. Round trips get shorter, raw data is processed where it is created, and a site keeps working when its link to the internet doesn't.
XcellHost supplies, installs and runs those nodes as one fleet. Each node runs VMs or containers you deploy from a single console; it connects back to XcellHost's Mumbai cloud over an encrypted link for control, backup, logs and the workloads that belong in the core. Our NOC patches and watches every node 24×7, and you get one INR invoice for the lot.
Many sites, real-time work, thin or patchy links.
If any two of these describe you, part of your stack belongs at the edge.
Billing, inventory and kiosks in hundreds of stores that must sell even when the broadband is down.
POS · price sync · footfallMachine data, quality vision and MES that react in milliseconds on the shop floor.
OT data · vision · MESPACS images, bedside devices and HIS that cannot wait on a distant region.
PACS · HIS · devicesBranch servers, queue systems and CCTV analytics with data that stays in India.
branch apps · CCTVScanners, sorters and yard cameras running at hubs with variable connectivity.
WMS · scanning · ANPRGame servers, live encoding and venue apps placed close to players and fans.
game servers · encodingEverything a site needs to run on its own — and everything you need to run 200 of them.
Right-sized servers we ship to your site, or capacity in a partner metro data centre*. Each node comes racked, imaged, hardened and enrolled in your fleet before it is switched on.
- S, M and L node sizes; GPU on quote
- Zero-touch provisioning on first boot
- Hardware swap under the managed plan
Run a Windows VM for the store app beside containers on lightweight Kubernetes — one node, both models.
When the WAN drops, the node keeps serving and queues data locally, then syncs in order once the link returns.
Deploy from Git or the console; releases roll out site by site with health checks and one-click rollback.
Encrypted disks and backhaul, node firewall, signed images and no inbound ports open to the internet.
Health, capacity and security alerts flow to XcellHost's Mumbai NOC and SOC; patches go out in agreed windows.
NVMe on every node for databases, video buffers and cached content, backed up to Mumbai on a schedule.
GPU-equipped nodes* run vision and language models beside cameras and machines, so frames never leave the premises.
Decide at the edge. Control, keep and report from Mumbai.
Live traffic stays local; control, backups, logs and heavy analytics travel over one encrypted link to the core.
Cameras, terminals and machines reach the edge node over the site LAN — typically a millisecond or two away.
Apps answer, models infer and data is filtered on site. The node queues anything bound for the core.
Events, summaries, logs and backups cross the encrypted link — not the raw video or every sensor tick.
Releases and policies flow down from the control plane; our NOC and SOC watch every node 24×7.
Pick the tier per workload — mix all three in one fleet.
A node in your store, plant, hospital or branch, on your LAN. Best for POS, machine control, CCTV analytics and anything that must survive a WAN outage.
- Anywhere in India we can ship and install
- Your site power & UPS; we size the node
- Runs offline, syncs on reconnect
Capacity in partner data centres* in major metros, close to your users and to the city's carriers. Best for user-facing apps, game servers and regional APIs.
- Delhi NCR · Bengaluru · Chennai · Hyderabad · Kolkata · Pune · Ahmedabad (partner data centres*)
- Multi-carrier connectivity per facility*
XcellHost's Mumbai cloud in a Tier IV data centre — the fleet's control plane and the home for central databases, backup, analytics and log retention.
- ISO 27001 · 27017 · 27018 · SOC 2 Type II
- 24×7 NOC & SOC
- Backups and logs kept in India
* Only Mumbai is an XcellHost facility. Metro edge locations are partner data centres* — availability per city is confirmed on your quotation. Latency bands are illustrative; actual figures depend on your links and carriers.
Data processed where it's created — and evidence kept where auditors expect it.
Faces in CCTV, patient images and customer data can be processed on site and discarded, with only events sent onward — data minimisation by design. Helps you meet your duties as a data fiduciary; your own notices and consents still apply.
Node and application logs are shipped to Mumbai and retained for 180 days in India, with clocks synced to NTP. Our SOC helps you report incidents within the 6-hour window.
For regulated entities, payment and policy data stays on Indian soil — at the site, in partner data centres* in India, or in Mumbai. Helps you meet localisation and outsourcing expectations.
XcellHost is MeitY empanelled and certified to ISO 9001, 27001, 27017, 27018 and 20000-1, with SOC 2 Type I & II reports for the core.
"Helps you meet" — compliance also depends on your own policies and controls. We map edge controls to your obligations with your compliance team.
What runs at the edge, and what still goes to Mumbai.
Stores that sell through an outage
POS, inventory and price look-ups run on the store's node. Footfall and shelf cameras are analysed on site. Sales, stock and events sync to Mumbai for head-office reporting.
Shop-floor decisions in milliseconds
Machine data is collected over OPC UA or MQTT, quality cameras are inspected by a model on a GPU node, and alerts fire locally. Daily summaries feed ERP and analytics in Mumbai.
Imaging and bedside data, on the premises
PACS caches and modality workflows run beside the scanners; bedside device feeds are buffered on site. Reports and archives move to Mumbai on schedule.
Branches with fewer boxes to manage
One managed node replaces scattered branch servers for queue systems, local apps and CCTV analytics. Transactions and evidence are kept in India.
Hubs that keep scanning
WMS look-ups, sorter control and ANPR at the yard gate run on the hub's node, so a flaky link doesn't stop the dock. Movements sync to the central system.
Closer to players and viewers
Game servers and match-making in metro edge locations*, live encoding at the venue, and origin storage in Mumbai. Pair with CDN for delivery at scale.
Priced per edge node, in rupees, every month.
Indicative · excl. 18% GST — confirm with XcellHost price list. Hardware, imaging, OS, monitoring and patching included in each node price.
8 vCPU · 32 GB RAM · 1 TB NVMe
- A store, kiosk cluster or small branch
- VMs or containers
- 24×7 monitoring & patching
16 vCPU · 64 GB RAM · 2 TB NVMe
- A plant line, hospital wing or hub
- Video analytics on CPU
- Offline queue & sync
32 vCPU · 128 GB RAM · 4 TB NVMe
- A whole plant or metro edge app
- Clustered pairs for HA
- Large local cache
GPU nodes · HA pairs · app operations
- GPU nodes for AI inference
- Application-level operations
- Site survey & design
Configure your edge fleet
Managed edge is priced on quotation and added to your quote request.
Indicative · excl. 18% GST — confirm with XcellHost price list. Site installation, power and local network are quoted per site.
See the difference before you buy a node.
Latency map
Pick where your users are. We compare a round trip to the Mumbai core with the nearest edge — illustrative estimates from distance, not measurements.
Backhaul saver
Cameras at each site, filtered at the edge. See how much video you no longer push to the cloud — and the uplink each site needs.
Illustrative: continuous streams, 30-day month, decimal TB. Actual savings depend on your cameras, codecs and the events you keep.
Central cloud only, a server room per site, or a managed edge?
| Central cloud only | Server at every site | XcellCloud Edge | |
|---|---|---|---|
| Response for on-site users | Tens of ms per round trip to a far region | ~1–2 ms on the LAN | ~1–2 ms on site, ~3–10 ms metro* |
| When the WAN fails | Site stops working | Keeps running, no central view | Keeps running, syncs on reconnect |
| Backhaul bandwidth | Every frame and tick goes up | Low, but data stuck in silos | Only events, logs & backups |
| Rolling out an update | Once, centrally | Site by site, often by hand | One push, staged, with rollback |
| Patching & monitoring | Provider's platform, your VMs | Your IT team, every site | XcellHost NOC, 24×7 |
| Cost model | Pay-as-you-go plus bandwidth | Capex hardware every 4–5 years | Per node per month, hardware included |
| Personal data exposure | Raw data leaves every site | Stays on site, often unencrypted | Processed on site, encrypted, minimised |
From site survey to a live fleet — in five steps.
We map sites, links, latency and the workloads that should move to the edge.
week 1Node size per site, backhaul, HA needs and the split between edge and Mumbai core.
week 2One or two sites go live first; we tune releases, offline behaviour and alerts.
weeks 3–4Zero-touch nodes ship to each site; your apps deploy from the console in waves.
per wave24×7 monitoring, patching, hardware swaps and monthly fleet reports.
ongoingTimelines are typical for a pilot of 1–2 sites and depend on site readiness and hardware lead times.