Test the network yourself
Ping and download test from the Baku data centre, before you buy.
ping lg.jezahost.comwget http://lg.jezahost.com/100MB.binwget http://lg.jezahost.com/1GB.bin
This is an ordinary customer VM (4 vCPU), not a dedicated test box.
How to test
- ping · shows latency: how close we are to you
- wget · real download speed; it measures the route, not the port
- Compare your result with the table below
Sample output: typical result from Baku; yours will differ with distance.
Public status page
Real-time state of every service and the uptime history.
Last 90 days
- VPS platform
- 100%
- Portal
- 99.97%
- 99.98%
Measured by Better Stack; planned maintenance is not counted as downtime.
Measured latency
Average ping, ms (check-host.net, 29 September 2026).
- Moscow40
- Tehran47
- Chișinău50
- Istanbul50
- Stockholm59
- Belgrade66
- Warsaw69
- Frankfurt71
- Amsterdam75
- London76
- Milan80
- Karaganda82
- Kyiv94
- New York143
- Singapore234
Measured disk performance
fio, 4 vCPU VM, direct I/O, libaio, 8 GB file, 2-copy replication included, 29 September 2026.
- 4k random read
- ~200,000 IOPS
- 4k random 70/30
- ~190,000 IOPS
- 134k read + 57k write
- Sequential read / write
- 2.9 / 2.2 GB/s
NVMe cluster: measured figures, not a guarantee.
Also measured on the smallest configuration (2 vCPU / 2 GB, different node): ~196k random read IOPS, 3.0 / 1.9 GB/s sequential. Disk performance does not depend on VM size.
fio --name=randread --ioengine=libaio --size=8G --rw=randread --bs=4k --iodepth=32 --numjobs=4 --direct=1 --runtime=60 --time_based --group_reportingfio --name=randrw --ioengine=libaio --size=8G --rw=randrw --rwmixread=70 --bs=4k --iodepth=32 --numjobs=4 --direct=1 --runtime=60 --time_based --group_reportingfio --name=seqread --ioengine=libaio --size=2G --rw=read --bs=1M --iodepth=8 --numjobs=1 --direct=1 --runtime=20 --time_basedfio --name=seqwrite --ioengine=libaio --size=2G --rw=write --bs=1M --iodepth=8 --numjobs=1 --direct=1 --runtime=20 --time_based
Network facts
- Port
- up to 1 Gbit/s
- IPv4
- Free static address, yours alone (how it works below)
- Traffic
- 1 TB per vCPU included
- IPv6
- One public address per server, free, at checkout or from the portal
- DDoS
- Network traffic is monitored. During an attack, only the targeted address is temporarily restricted; other customers are not affected.
- Isolation
- A separate isolated network per customer
- BGP
- BGP session: announce your own AS and IP blocks from your VPS (IPv4 and IPv6, RPKI-validated).
- Send a request Learn more
Network and security
Firewall included, free
Your server is visible from the internet only on the ports you open. Everything is closed by default. You manage the rules yourself in the portal, and they apply instantly.
You choose who can connect
You choose who can connect: everyone, Azerbaijan, Türkiye, Russia, Georgia, Uzbekistan, Cloudflare or your own IP list. If you need a country that is not on the list, write to support and we will add it. The default follows your account's country: «Azerbaijan only» for Azerbaijan, «All countries» for any other country.
One IP: one server or several, your choice
Forward any port or port range to any of your servers: give the whole IP to one server, or split it across several. A range counts as one rule.
You choose the outbound address
Your servers go out to the internet from the IP you choose: one address for the whole account, or a separate outbound address for each server (e.g. a mail server on its own additional IP).
Trusted addresses
Add your office or home IPs (single addresses or ranges) to the trusted list, and open sensitive ports such as RDP, SSH or a database only to them. The list is common to all your rules: change it and the rules update themselves.
What's different
Typical VPS
- The IP is bound directly to the server
- All ports are open
- You set up the firewall yourself
jezahost
- Everything is closed by default
- You open ports from the portal
- You split one IP however you like
- Want a fully open server? Forward all of the IP's ports to one server with 2 rules. For RDP you pick another external port instead of 3389. It keeps automated attacks away.
- Every account gets a free static IPv4 that is yours alone (it is not shared with other customers).
- Additional static IPs: as many as you need; to keep different services on separate addresses or open the same port (e.g. 443) on two servers. 5,00 AZN (~$2.94) per month.
Our network
The IPv4 (95.86.162.0/24) and IPv6 (2a00:c0a0:8000::/34) blocks are assigned in RIPE NCC to JEZA CLOUD LLC (ORG-JCL8-RIPE) and announced over BGP from our own AS (AS218452).
Autonomous system
AS218452
Registered with RIPE NCC to JEZA CLOUD LLC.
IPv4 block
95.86.162.0/24
Assigned in RIPE NCC to JEZA CLOUD LLC (ORG-JCL8-RIPE).
IPv6 block
2a00:c0a0:8000::/34
Assigned in RIPE NCC to JEZA CLOUD LLC (ORG-JCL8-RIPE).
Connectivity
BGP
We announce both blocks from our own AS over BGP.
RPKI: all prefixes valid (ROA)
View on bgp.toolsTest our network yourself
You don't have to take our word for it. Two public measurement nodes run on our network in Baku (AS218452). Use them to test our network from the outside, or run ping, traceroute, MTR, HTTP and DNS tests from Baku to any address.
In the results, the “Azerbaijan, Baku” node is ours. It is currently the only Check-Host node in Azerbaijan.
check-host.netHardware, data centre and reliability
Hardware
- Processor
- Intel Xeon Platinum 8268
- Memory
- DDR4 ECC
- Disk
- Enterprise NVMe: two copies, on two different physical servers
Data centre (Baku)
- 2 generators
- 2 UPS units
- Internet from 2 independent backbones
- 2 professional cooling systems
How reliability is built
- Monitoring never stops: physical servers, disks and the network are watched automatically.
- The platform is a cluster of several physical servers (nodes) in one data centre.
- If a node fails, the VMs on it start automatically on another node of the cluster, without human intervention. In testing this took about 3 minutes; not a guarantee.
- Every VM's disk is kept in two copies, on two different nodes.
- Only disaster-recovery backups are sent off-site. There is no second running system there. If the whole data centre were lost, recovery would be made from those backups and would take time.
Support
Every day, weekends included, 08:00–24:00 (Baku time). Target response time: 2 hours during support hours; a request that arrives outside them, 2 hours after 08:00 the next day. Reach us via a ticket in the portal or [email protected].
Service level agreement (SLA)