Nodes are devices running pplznet firmware. Nodes are interconnected to each other via peerings.
The general color for Nodes is 7EA6E0 as long as it's not defined otherwise.
The term clusternode is being used if one talks about nodes without being specific about the exact type of node. It can be any type of node running within a cluster.
An Isonode has zero (=0) active peerings. This type of node should never exist because it's isolated and cannot deliver any purpose. Please contact your cluster board for help.
A Solonode has exactly one (=1) active peering partner. In an ideal environment Solonodes are rare to non existant. It should only be operated where no secondary peering is possible. Two peerings to the same neighbour usually does not make sense. For questions contact your cluster board.
A so called Duonode has exactly two (=2) peering partners. This the most common mode as the cluster should be made of interconnected rings. It provides failover to the cluster.
Nodes that have more than two (>2) peering partners are called Polynodes or Multinodes.
The rule bigger is better does not apply in a pplznet cluster. Too many peerings poses a potential problem. It will decrease the maximum size of a cluster and in tendency makes the cluster slower. If in doubt, contact your cluster board.
As 2 peerings is the best option for most scenarios, nodes with 3 or more peerings are desirable when they are in the core of the cluster. Especially around supernodes it's desirable to have an amound of 3 or more peerings per node.
Contact the cluster board for further questions.
A node connecting clusters is called an Internode. Internodes usually come in pairs, one in each cluster. Sometimes Internodes have more than one peering. Multiple Clusternodes should be connected to the internode in order to provide fault tolerance and optimal performance. There can be more than on peering between two clusters.
The Link between two Internodes is called an Interlink.
The colour code for Internodes is FF66FF.
A supernode is a dedicated Router / Gateway that delivers Internet connectivity to the cluster and its members. Funding of those supernodes is done via the shares of their users. The share varies and is defined by the cluster board.
The absolute minimum amount of peerings to other nodes should be 2, better 3 to 5. The supernodes peering go into most different geographical directions.
The colour code for Supernodes is FF0000.
An Uplinknode has at least one peering to a Supernode. This type of peering is called a Superlink. As Uplinknodes are important for the availability of the Internet Uplink for the Cluster they should have more than two (>2) peerings to other Nodes, which are no Supernodes.
A Fosternode is a node that runs within a cluster that for management purposes. This might for example be a node for monitoring, deployment or central DNS purposes. Those nodes are being operated by either Cluster Operators or a even higher administrative instance.
As Foster nodes are important they should have good connectivity, meaning their peering partner shall not be a Solonode. Depending on what the Fosternode offers there shall be a concept how to achieve High Availability for this service.
The colour code for Fosternodes is 7EA6E0.
A Roguenode is a node that tries to destroy or interfere with normal cluster operations. It might for example try to advertise a false route or to flood the cluster in order to interrupt the service.
With good monitoring and other mechanisms the problem of Roguenodes shall be minimized.
An Anarchynode is not part of the managed cluster but is unmanaged and doesn't fit perfectly as it poses a risk to a trouble-free cluster operation.
An Anarchynode might work perfectly and pose zero problems, but as it is a potential risk one should rather address the cluster board in order to go from an Anarchynode to a regular one.
When neighbours interconnect their nodes via peerings, a cluster of nodes is being formed. The amount of nodes within a cluster is limited so typically a city needs several clusters.
A stack is the combination of two or more (>1) clusters. Clusters are being interconnected so they can communicate with each other. In order to have no IP conflicts the IP adressing is being managed by the pplznet Founders.
Someone who runs a node is a node owner. The owner typically is not the operator as the operation is done by the cluster board and its Cluster Operators. The owner's node is usually located at his home.
The responsible administrator for a node is called Node Operator.
Planning and supervision of a cluster is being done by a Cluster Operator. A Node Operator may at the same time also be a Cluster Operator.
Cluster Operators are responsible for the cluster to be in optimal shape. They monitor size and topology of it. If Node Operators need technical assistance their first contact is the Cluster Operators.
Cluster Operators are being determined by the Cluster Board.
The operators of Supernodes are being called Supernode Operators. They are determined by the pplznet Founders.
The operator of Internodes are being called Internode Operators. They are determined by the Cluster Board they reside in.
A cluster needs a Cluster Board. It makes operational decisions within their own cluster.
The Stack Board is the same for the board as the Cluster Board is for a cluster. Every Cluster Board selects a representative for the Stack board.
The Founders of pplznet act to all pplznet Clusters and Stacks as the IANA acts to the Internet. They are responsible for main development and also for handing out IP subnets to new clusters. Strategic decision are made by the pplznet Founds.