Here you will find an overview of important network-related topics. gridscale offers the use of public IPs as well as the use of internal networks.
Each server can be assigned both an IPv4 and an IPv6 address. You can either select existing ones or create a new one with one click. IP addresses can be connected either directly in the creation process or afterwards.
All purchased public IPs overview are visible in the “IP-Manager”.
A public IP is assigned to a connected Server via DHCP. You can also assign the IP in a static mode but you have to choose the correct WAN Interface on your server.
Note: For security reasons we drop all traffic on ports 111,135,137,138,139,445. If you would like to interact with external services using these ports please use a VPN.
IP and Server Network Ordering
By default the first connected network will be your WAN interface in the server. If you need your WAN on another interface (mostly for Firewalls as there is a management interface on the first one) you have to connect the private network first and the public one afterwards.
A click on the network tile shows the MAC addresses, which are numerically ordered.
Multiple Public IP Addresses
In case you need more than one IP on a server you can assign a second one via “IP-Manager” (or choose an existing one) and you have to set this one to “Failover IP”. Setting an IP address as a failover enables you to use it to manually configure a group of redundancy within your infrastructure, or simply allow you to transfer traffic between configured servers with ease. Failover IPs need a manual configuration on your server (e.g. Alias) through the interface.
We do not allow CARP on our Public Network.
We give you the freedom to use servers without external IPs - for example, servers that should only be accessible in an internal network. A Private Network functions as a Layer 2 network.
Networking Between Accounts
Private Networks cannot span over projects and other public accounts. If you have to setup an internal network over projects you can upgrade to our Partner Panel and use the extended Software-Defined Network (SDN) feature.
There is no option to use VLAN in our private networks. In case you need this for any reason you can setup a VLAN on your server’s network interface.
VLAN is supported by our SDN, which is explicitly available by upgrading to our Partner Panel.
To receive a private IP Address, you will need to enable DHCP on your private network.
DHCP is also available for Virtual Switches and should be configured by the Partner.
By default, the DHCP server provides IP addresses in ascending order to servers connected to the private network.
The default network range is 192.168.121.0/24 which can be customised via the API, with future support coming to the Cloud Panel soon.
Warning: changing the DHCP range manually will override any previously saved confiration and IP leases if they’re not within the new range.
Once a server is connected, the servers private IP Address can be seen via the network relation via the API, or via the network relations within the Panel.
Turning off DHCP, will not reset the configuration, and the server will keep it’s private IP on the network interface until restarted.
Note: if you have a server connected to multiple DHCP networks, make sure to customise the IP range served via DHCP, servers connected to multiple networks with overlapping ranges can cause unintended network issues.
It is possible to specify a DNS or Gateway Server IP Address via the API. We require these addresses to be within the defined DHCP range. We do not provide DNS or Gateway functionality, but reserve the addresses for you to provide these services if you wish. Just fix the IPs to your DHCP option servers connected to the network.
These IP addresses must reside within the networks IP range.
Subnets can be reserved via the api. Keep a few things in mind when reserving a subnet:
- The subnet must exist within the IP range: the default is 192.168.121.0/24.
- The subnet cannot confict with DNS/Gateway addresses given.
- The subnet should not conflict with any IP which you have fixed to a server.
Fixing IP Addresses
Once a server is connected to a private network with DHCP enabled, it’ll dynamically receive an IP address. If you would like a specific IP address to be fixed to that server, you can do so via the API.
Fixed IPs have to be within the network range, and have to be outside of the reserved subnets or reserved gateway/DNS addresses.
How do I request or delete IP addresses?
You can manage your IP addresses within the IP Manager. Typically, deleting a server from other providers results in losing access to your IP addresses at the same time. We have decided to separate these objects so that you have the greatest possible flexibility.
What are failover IPs?
If you start a server, then you can specify the IP addresses and we configure the network interface automatically for you. However, there are situations where you may want to take control of the configurations. A possible case is using an IP address between multiple servers without talking to the API. By setting an IP address to failover, we do not interfere with the configuration and unlock the IP addresses for all your servers. This way you can set them up manually in the network configuration of your server. This allows you to create and operate high availability solutions in the same way as in a physical data center, replicating real-world scenarios into the cloud.
How do I connect servers in the internal network?
We do not intervene in the connection of servers in the internal network. You can freely define IPs or VLAN tags through this transparent Layer 2 connection.
What is the difference between public and private networks?
Within the public network, IP or MAC addresses that do not belong to you are restricted from being used. Multicast and ARP will be filtered as well. The public network supports an MTU of 1500, whereas in private networks there are no filters even for MAC addresses. The private networks offers Layer 2 domains with an MTU of up to 9000.
You can create as many private networks as you like, but there is a limitation to a maximum of 8 connected networks per server (the public network also counts here). Each network connected to the server is its own virtual network card with its own MAC address.
Does the size of the server affect network performance?
No, there aren’t any dynamic limitations. You always get the full performance of the components used.
Does each server get a fixed external IP?
You can assign both IPv4 and IPv6 to each server when you create it or later in the server element itself. But we also give you the freedom to use servers without external IPs - for example, servers that should only be accessible in an internal network. If you need more than one IPv4/IPv6 address on a server, you can add additional IPs in the IP Manager and mark them as a failover IPs. These IPs can then be manually configured on each of your servers.
Can I assign a second IP address to my server?
Each server can have a maximum of one automatically configured IPv4 and one automatically configured IPv6 address. However, you can configure any number of IP addresses in the IP Manager to failover and then manually set them in your server’s network configuration. With a failover IP the configuration remains unchanged and the IP addresses are activated for all your servers.