To plan for the distribution optimization in your peer-to-peer deployment, consider how data transfers within networks.
Offices and Subnets
Define an office with one or more subnets or subnet ranges connected over a local area network (LAN). Offices retrieve the content from their parent offices, and distribute them to their child offices.
Peer distribution has three types of offices, and these office types share data in specific ways.
- Default - Defines a standard wired LAN. Clients attempt to the share content and they send broadcast discovery requests.
- VPN - Defines an office and subnet range allocated for clients connecting through VPN. Clients within a VPN office do not attempt to the share content, but they do send broadcast discovery requests.
WiFi - Defines an office and subnet range allocated to clients connected over WiFi. Clients within a WiFi office share content, but they do not send broadcast discovery requests.
If you have a physical office with a wired (default) subnet and a WiFi subnet, create an office for each network. Make the WiFi office a child of the wired office so that the WiFi network receives packages from the wired parent office.
Central Office and the Peer-to-Peer Server
The peer-to-peer server must reside in one of the subnets in the top-tiered Central Office. This placement makes it available to all clients in the hierarchy.
Data Transport in Offices
The system distributes content from a parent to child office once. This behavior limits data sent across wide area network (WAN) links.
The adaptive protocol is a proprietary protocol that monitors the length of edge router queues and sends data when queues are nearly empty. This protocol, implemented by an advanced kernel driver, removes the need to throttle the bandwidth when deploying applications with the peer distribution.
Data transport within offices uses the LAN, or Foreground protocol. The peer distribution system does not manage this protocol.
Data transport between offices uses the WAN, or Background protocol. This protocol is also called the Adaptive Protocol that protects the bandwidth availability on WAN links.
Define subnets connected over a WAN link as separate offices. If offices are misconfigured, the LAN protocol might be used over a WAN link, causing saturation of the WAN.
Clients Receive Applications According to Ordered Criteria
The peer-to-peer system sends and receives applications according to many factors, including the available device space, device form factor, and operating system type. The download order follows these elections from top to bottom.
- Devices with the largest actual free space
- Devices that are identified as preferred, also called RVPs (rendezvous points)
- Device chassis type (desktops are selected over laptops)
- Device operating system type (servers are selected over work stations)
- Devices with the longer system up-times
- Devices with the largest usable free space
Peer-to-peer clients receive application packages from a CDN or a file storage system when they cannot find packages within the hierarchy. A CDN, which is optional for on-premises deployments, offers increased download speed over the file storage system.