Dedicated Servers
Dedicated Servers
One physical machine, one tenant, and an encrypted overlay network fitted as standard so it behaves as though it is sitting on your own LAN.
A dedicated server is one physical machine with one tenant on it: yours. No hypervisor taking a cut, no neighbour on the same socket, and the whole of the memory bandwidth, the NVMe queue depth and the network card available to your workload. Every machine ships joined to our encrypted overlay network, so your offices, your existing servers and the new hardware all sit on the same private address space and reach each other as if they were patched into the same switch. Public and private networking are separate ports, and the private side runs at 25 Gbps or 50 Gbps depending on the range.
Who It Is For
- Databases and workloads that cannot share a socket
- Teams consolidating an ageing on-premises rack
- Businesses replacing a cloud bill that scales faster than the traffic
What It Is
A single-tenant AMD EPYC machine with no hypervisor between your workload and the hardware
How It Reaches You
Joined to your encrypted overlay network on build, so offices and existing servers see it as a local machine
How It Is Billed
A fixed monthly figure per machine, plus a one-off installation charge quoted up front
Who Answers
The engineer who specified and racked it, with a managed layer available
Specification
Processors
AMD EPYC, 8 to 24 physical cores, up to 5.7 GHz boost
Memory
64 GB to 3 TB DDR5 ECC, depending on the range
Storage
NVMe from 2 x 960 GB up to 6 x 15.36 TB
Public Network
1 Gbps guaranteed, burst to 5 Gbps or 25 Gbps
Private Network
25 Gbps or 50 Gbps on a separate port
Overlay Network
Fitted by default, so the server behaves as if it is on your LAN
What You Get
The Whole Machine, Not A Share Of One
There is no hypervisor between your workload and the silicon, so the cores, the memory channels, the NVMe queues and the network card are all yours. Workloads that are sensitive to jitter, such as databases and build farms, behave the same at nine in the morning as they do at midnight.
On Your LAN From The First Boot
Every server joins your encrypted overlay network as part of the build, not as a later project. Your branch offices, your existing on-premises kit and the new hardware share one private address space, so file shares, domain traffic and internal APIs work exactly as they would across a patch panel. There is no VPN concentrator to size and no public port to expose.
Storage Sized To The Job
Ranges start at a mirrored pair of 960 GB NVMe drives for a database or an application server, and run to six 15.36 TB drives where the workload is an archive, a backup target or a Ceph node. RAID level and filesystem are decided with you rather than fixed by a catalogue.
Operated By The Team That Racked It
Provisioning, firmware, disk replacement and capacity review sit with the same engineers who specified the machine. Take it bare and run it yourself, or hand over monitoring, patching and backup verification and keep one number to call.
Pricing
Four standard machines, priced per month with a one-off installation charge. Every one of them ships joined to your encrypted overlay network, so the server can be used as if it were sitting on your own LAN alongside the kit you already run. Memory and storage are ranges: the figure shown is the entry configuration for that model.
INDEDI-1
₹38,570
Excluding GST, per month
Installation ₹38,570, one-off
- AMD EPYC 4345P, 8 cores and 16 threads
- 3.8 GHz base, 5.5 GHz boost
- 64 GB to 256 GB RAM
- 2 x 960 GB NVMe, up to 2 x 960 GB plus 2 x 15.36 TB
- Public 1 Gbps, burst to 5 Gbps
- Private 25 Gbps
- Overlay network included, use it as if it is on your LAN
INDEDI-2
₹57,855
Excluding GST, per month
Installation ₹57,855, one-off
- AMD EPYC 4585PX, 16 cores and 32 threads
- 4.3 GHz base, 5.7 GHz boost
- 64 GB to 256 GB RAM
- 2 x 960 GB NVMe, up to 2 x 960 GB plus 2 x 15.36 TB
- Public 1 Gbps, burst to 5 Gbps
- Private 25 Gbps
- Overlay network included, use it as if it is on your LAN
INDEDI-3
₹1,15,615
Excluding GST, per month
Installation ₹1,15,615, one-off
- AMD EPYC 9135, 16 cores and 32 threads
- 3.65 GHz base, 4.3 GHz boost
- 128 GB to 3 TB RAM
- 2 x 1.92 TB NVMe, up to 6 x 15.36 TB
- Public 1 Gbps, burst to 25 Gbps
- Private 50 Gbps
- Overlay network included, use it as if it is on your LAN
INDEDI-4
₹1,21,410
Excluding GST, per month
Installation ₹1,21,410, one-off
- AMD EPYC 9255, 24 cores and 48 threads
- 3.2 GHz base, 4.3 GHz boost
- 128 GB to 3 TB RAM
- 2 x 1.92 TB NVMe, up to 6 x 15.36 TB
- Public 1 Gbps, burst to 25 Gbps
- Private 50 Gbps
- Overlay network included, use it as if it is on your LAN
Figures are monthly and exclude GST, and the installation charge is one-off. Because the overlay network is fitted as standard, the machine can be used exactly as if it were on your LAN: the same private address space as your offices and your existing servers, with no VPN concentrator and no public port opened to reach it. Memory upgrades, additional NVMe, extra public addressing, Windows licensing and the managed operations layer are quoted on top.
How It Works
Five stages from an enquiry to a machine that is already on your network. Nothing here needs a call to get moving.
Sizing
We start from the workload rather than a model list: core count against the way the application actually threads, memory against the working set, and NVMe against the read and write pattern. The output is a written specification naming a model, a storage layout and a monthly figure.
Build And Rack
The machine is racked, firmware is brought to a known baseline, and the storage is laid out with the RAID level and filesystem agreed in the specification. Out-of-band management is configured at the same time, so console access does not depend on the operating system being healthy.
Join The Overlay
Before handover the server joins your encrypted overlay network, so it comes up already reachable from your offices and your existing servers on one private address space. From that point it can be used as if it were a machine on your own LAN, without a VPN concentrator and without a public port opened to reach it.
Handover
Root credentials, out-of-band management and the overlay configuration are handed over together, with the address plan written down. Nothing is held back behind a support ticket.
Operate
Monitoring, firmware, disk replacement and capacity review either stay with your team or come to us as a managed layer. Failed drives are replaced by us either way, because the hardware is in our rack.
Against Public Cloud Instances
Most enquiries for bare metal arrive from a cloud bill rather than from a smaller server. These are the differences that change both the invoice and the way the machine behaves.
Tenancy
- Public Cloud Instance
- A slice of a host shared with tenants you cannot see, behind a hypervisor
- Our Dedicated Servers
- One physical machine with one tenant on it, and no hypervisor in the path
Performance Consistency
- Public Cloud Instance
- Steal time and burst credits, so the same query is not the same query at peak
- Our Dedicated Servers
- Full memory bandwidth, full NVMe queue depth and a network card nobody else is using
Reaching It Privately
- Public Cloud Instance
- A gateway, a peering arrangement or a VPN appliance to size, licence and keep patched
- Our Dedicated Servers
- An encrypted overlay fitted on build, so it answers as if it were on your LAN
Egress
- Public Cloud Instance
- Metered per gigabyte, so the bill moves with traffic you do not control
- Our Dedicated Servers
- A guaranteed public port with a burst ceiling, quoted in the monthly figure
Forecasting
- Public Cloud Instance
- A figure you reconcile at the end of the month against usage lines
- Our Dedicated Servers
- The same figure every month, agreed before the machine is racked
Who Answers
- Public Cloud Instance
- A ticket queue and a status page written for every customer at once
- Our Dedicated Servers
- The engineer who specified the machine and put it in the rack
What We Need From You
Five things, and a specification comes back. None of them require a meeting first.
A Workload Description
What the machine runs and how it threads. A database that wants clock speed and a build farm that wants core count point at different models in the range.
A Storage Position
Working set today, growth over the next two years, and whether the pattern is read heavy, write heavy or archival. That decides drive count, RAID level and filesystem.
Your Network Plan
The private ranges already in use across your offices and existing servers, so the overlay address space is allocated without colliding with anything you run.
Availability Expectations
How long the workload can be down for a firmware update or a drive replacement, and whether a second machine is needed to make that window acceptable.
A Named Contact
One person who can approve the specification and take the credentials at handover, so the build does not stall waiting on a signature.
Send those five and a written specification comes back naming a model, a storage layout and a monthly figure. An engineer picks it up, not a form.
Start An EnquiryQuestions About Dedicated Servers
The technical ones that come up before anyone asks for a figure.
One encrypted private address space spanning your offices, your existing on-premises servers and the machines in our rack. Traffic between them is peer to peer and encrypted end to end, so the dedicated server can be used exactly as if it were on your LAN: domain traffic, file shares and internal APIs all work without a VPN concentrator to size, and without a single management port exposed to the public internet.
No. It is fitted as part of the build and included in the monthly figure. The machine is handed over already reachable from your network rather than left for a second project to connect.
Yes, within the range the model supports. The advance machines take up to 256 GB of memory and up to two 15.36 TB drives alongside the boot pair; the scale machines take up to 3 TB of memory and six drives. An upgrade is a scheduled window and a revised monthly figure, not a migration.
We replace it, because the hardware is in our rack. On a mirrored or parity layout the machine stays up through the replacement, and we tell you before we touch it rather than after. Nothing about the process depends on you holding spares.
Yes. A common pattern is Karl on the dedicated machine with your own workloads on top, which keeps the hardware single tenant while you divide it up as you like. We will lay the storage out for that at build time if you tell us in the specification.
Either, and you can change your mind later. Taken unmanaged you get root, out-of-band management and the overlay configuration, and we stay out of the operating system. Taken managed, monitoring, patching and backup verification move to us, with the same engineers on it.
Specify A Machine With Us
Send the workload and the growth you expect from it. We will come back with a model, a storage layout and a monthly figure, with the overlay network already in the scope.
Sales
- New servers, identity rollouts and workstation pilots
- Specifications, sizing and monthly pricing
- Moving an existing estate across to us
sales@infinite-networks.net
Support
- Anything already running with us
- Incidents, capacity changes and restores
- Accounts, access and policy changes
support@infinite-networks.net
Office
#002, Ceaser's CastleSainikpuri, Defence Colony
Hyderabad 500 094
India