Where exactly are the servers?

In the Equinix IL4 data center in Ümraniye, Istanbul, in racks owned and operated by CloudBT.

Is the hardware yours?

Yes. HPE and Dell servers and Arista and Juniper network equipment owned by CloudBT. We do not resell another provider's virtual servers.

Are resources really dedicated?

On VPS (VDS) plans the vCPU, memory and storage in your plan are reserved for your server. Virtual data center pools are likewise reserved per project.

What is your latency to Turkish operators?

1.4 ms to Cloudflare Istanbul, 1.7 ms to TurkNet, 1.8 ms to Türk Telekom, 6.2 ms to Vodafone and 14.2 ms to Turkcell. For comparison, Frankfurt is 35 to 45 ms away.

How often do you re-measure latency?

Whenever the network changes (new uplinks, carriers or equipment) and at least periodically; the date of the current measurement is shown on this page.

Do you have other locations?

No. All servers are in Istanbul; that focus is the point of the product.

Equinix IL4 · Istanbul

Redundant at Every Layer
Server Infrastructure

Q: How did you measure latency?

On October 5, 2026, from our production server in Equinix IL4, Istanbul. ICMP values are the average of ten packets; TCP values are the connection setup time to port 443, lowest of six attempts. The commands are published on this page so you can repeat the measurement.

Two separate halls (A/B) at Equinix IL4
2×100G redundant network on two switches, RAID10 NVMe SSD
AMD EPYC processors, HPE and Dell servers

View Pricing Contact Us

Rated 4.8from 600+ reviews.

Infrastructure

Redundancy Is a Property of the Platform, Not of the Plan

From the smallest plan to the largest, every server runs on the same redundant infrastructure. You do not pay extra for it.

2×100G uplinks

Every Host on Two Switches

Each host connects to two separate switches with a 100G link each (LACP). If a switch or cable fails, traffic keeps flowing over the other link; you will not notice.

2data halls

Automatic Failover Across Two Halls

Your virtual server runs in an HA cluster spanning halls A and B of Equinix IL4. If a physical host, or a whole hall, goes down, it restarts in the other hall by itself within 1–2 minutes.

RAID10NVMe

Mirrored NVMe Storage

All-NVMe SSD in a RAID10 mirror, synchronously replicated between the two halls. A failed disk or hall means no data loss.

A+Bpower feeds

Dual Power Feeds

Two power supplies in every host, two separate feeds from the facility. UPS and generators are provided by Equinix.

0reboots

No Downtime for Maintenance

When a host needs maintenance, your server is moved to the other hall with Live Migration while it runs. No midnight maintenance notices.

99.9% SLA Commitment

Equinix IL4
ISO 27001 Data Center
KVKK Compliant

When something fails

What Happens When Things Go Wrong?

Four scenarios. Follow what happens to your server, step by step.

Switch Failure

One of the two switches goes down.

1
Switch A stops responding
Monitoring flags the link immediately.
2
The LACP bond falls back to Switch B
The switch-over takes milliseconds; sessions stay up.
3
Traffic flows over one link
Capacity is sufficient; your server runs at the same speed.
4
The switch is replaced on schedule
No customer is affected during the replacement.

Included in every plan

Hardware fails. You will not notice.

Two switches, mirrored disks, two power feeds; and your server is redundant across two separate data halls. Redundancy is not a separate package, it is the platform itself: from the smallest plan to the largest, every server runs in the same redundant cloud, at no extra charge.

Explore Our Infrastructure View Plans

All systems operational

Live data, last read 04:52 · Status page

Physical nodes: 1/1 online
Longest uptime: 4 gün 22 sa
VPN gateway: running
To Türk Telekom: 1.8 ms

Equinix IL4, Istanbul
ISO 27001 data center
KVKK compliant
99.9% SLA commitment
HPE and Dell servers
Arista and Juniper switches

Measured latency · October 5, 2026

Latency Measured From Our Network

Numbers, not adjectives. Measured from our production server in Equinix IL4, Istanbul; we re-measure when the network changes.

Cloudflare Istanbul: 1.4ms; ICMP (ping)
Türk Telekom: 1.8ms; ICMP (ping)
Vodafone Turkey: 6.2ms; TCP connect time
Turkcell: 14.2ms; TCP connect time

Networks and services in Turkey

From our server in Istanbul to the operator networks and to services everyone uses.

Networks and services in Turkey
Destination	Measured latency	Method
Cloudflare IstanbulICMP	1.4 ms	ICMP
TurkNet (ISP)TCP connect	1.7 ms	TCP connect
Türk Telekom (DNS server)ICMP	1.8 ms	ICMP
Borsa Istanbul (stock exchange)TCP connect	3.1 ms	TCP connect
Vodafone TurkeyTCP connect	6.2 ms	TCP connect
Google Public DNSICMP	10.2 ms	ICMP
e-Devlet (government portal)TCP connect	10.8 ms	TCP connect
TurkcellTCP connect	14.2 ms	TCP connect

From the same server to data centers abroad

The path a visitor in Turkey would travel if your server stood in one of these cities.

From the same server to data centers abroad
Destination	Measured latency
Istanbul (CloudBT) → Türk Telekom	1.8 ms
Istanbul → Milan	30 ms
Istanbul → Frankfurt	35-45 ms
Istanbul → Nuremberg and Falkenstein	37-40 ms
Istanbul → Amsterdam	40-51 ms
Istanbul → Warsaw	45 ms
Istanbul → London	47-52 ms
Istanbul → Paris	47-56 ms
Istanbul → Stockholm and Helsinki	56 ms
Istanbul → US East Coast	117-130 ms
Istanbul → Singapore	188-190 ms

Operator networks

1.8 ms to Türk Telekom, 1.7 ms to TurkNet, 6.2 ms to Vodafone, 14.2 ms to Turkcell.

Services people use

1.4 ms to Cloudflare Istanbul, 10.8 ms to e-Devlet, 3.1 ms to Borsa Istanbul.

Data centers abroad

Frankfurt 35 to 45 ms, Amsterdam 40 to 51 ms, London 47 to 52 ms, US East Coast 117 to 130 ms.

How we measured

Measured from our production server in Equinix IL4, Istanbul. ICMP values are the average of ten packets; TCP values are the connection setup time to port 443 (DNS resolution excluded, lowest of six attempts). Values vary with network conditions.

Comparison values were measured from the same server to the test servers of public cloud providers in each city. A visitor in Turkey reaching a server in that city travels roughly the same path.

The visitor's own home or mobile connection adds its latency on top. The same addition applies to a server abroad, so the gap stays the same.

Server ping test: repeat the measurement yourself

# ICMP: average of ten packets
ping -c 10 1.1.1.1

# TCP: connection setup time to port 443 (lowest of six attempts)
for i in 1 2 3 4 5 6; do curl -o /dev/null -s -w "%{time_connect}\n" https://www.turktelekom.com.tr/; done

Why We Publish Measurements

Every provider says "fast". We publish the numbers, the method and the commands so you can repeat the measurement from your own server. 1.8 ms to the Türk Telekom network and 1.4 ms to Cloudflare Istanbul (measured October 5, 2026).

About the Data Center

Infrastructure Features

Every layer selected from enterprise vendors; all servers in our own racks in Equinix IL4, Istanbul.

HPE and Dell Platforms
Enterprise server platforms with redundant power and ECC memory.
AMD EPYC
Dual processors per host; reserved cores per virtual server.
All-NVMe
No SATA tier; every plan runs on NVMe.
2×100G Uplinks
Two 100G links per host, to two separate switches.
Redundant Power and Cooling
Equinix IL4 provides redundant power paths and cooling; hosts have dual power supplies.
Monitoring 24/7
The platform is monitored around the clock; critical incidents are handled without waiting for a report.

AMD EPYC Processors
Dual AMD EPYC processors with high core counts in every host. Your virtual server runs on cores reserved for you, not oversold.
All-NVMe Storage
The entire storage layer is NVMe SSD. Database queries, builds and file transfers finish many times faster than on SATA SSD.
2×100G Redundant Network
Each host connects to two separate switches with a 100G link each. If a switch or link fails, traffic keeps flowing over the other without interruption.

Same price, different infrastructure

How We Differ From an Ordinary VPS

The difference is invisible while everything works. It shows when things go wrong.

Situation	Ordinary VPS	CloudBT
If the physical host fails	You wait until the hardware is repaired. That can take hours.	Your server restarts in the other hall within 1–2 minutes, by itself.
If a whole data hall goes down	Every server in that hall stops; your data depends on the disks there.	Your storage is synchronous in the other hall; your server starts there within 1–2 minutes.
If maintenance is needed	A midnight maintenance notice and a reboot.	Moved with Live Migration while running. No reboot.
If a switch or cable fails	With a single link your server drops off the network.	Every host is on two switches. Traffic flows over the other link.
If a disk fails	Depends on the RAID setup and your last backup.	The NVMe RAID10 mirror keeps going. The disk is hot-swapped.
If a power feed is lost	With a single power supply the server shuts down.	The second feed takes over. The server keeps running.

We moved our e-commerce site from a server abroad to Istanbul. Load times for visitors in Turkey dropped visibly, and the server stayed calm during campaign peaks. The support team walked us through every step of the migration.

Serkan Y.

Verified customer

Frequently Asked Questions

Questions about the infrastructure.

In the Equinix IL4 data center in Ümraniye, Istanbul, in racks owned and operated by CloudBT.

Services Running on This Infrastructure

Manage Your Servers from One Panel

Console, snapshots, firewall, reinstall and resource graphs; also through the API and MCP.

Redundant at Every Layer
Server Infrastructure