Distributing Data Between Depots

The process of distributing mass data (e.g. Software Packages or Software Sets) between individual depots of a site structure is referred to as distribution. The data is distributed time-controlled (scheduled polling intervals), from top level to bottom level (vectored) and from nod to nod (radiating). The design of the infrastructure allows you to specify the sites between which the files will be distributed and the direction in which the data is distributed. You can also define which mechanism (Push or Pull) will be used to distribute data to the neighboring site for each site individually. Additionally you can determine the transfer protocol you want to use for distribution; currently, DSM supports SMB (Server Message Block) and HTTP (Hyperlink Transport Protocol).

It is possible to control the network load by defining distribution time frames and polling intervals in which you can specify when data is distributed and with which intensity.  

Number, Distribution and Configuration of Distribution Services

The Distribution Service is used to copy data (mass data, ICDB). It is installed on the Management Point that has a site, region or ORG assigned to it. You can only set up one Distribution Service per site, region or ORG. During the copy process, the Distribution Service copies the files stepwise from the central depot on ORG level to the next child master depot and then to the local depots (if available).

Push / Pull Mechanism

The Push or Pull mechanism determines the direction in which two computers communicate. The Pull mechanism allows a client to access a server and to query data. The client retrieves new data and information independently from the server and opens a new connection each time (client pull). The Push mechanism, in contrast, involves a security step which prohibits external clients from accessing the server. In specified time intervals, the server "pushes" data to the client. The Distribution Service employs the two mechanisms as follows:

  • Pull mechanism: The Distribution Service is installed on the target and copies files from the source to the target.
    Example: A Distribution Service is installed in a specific site; it copies the files from the central depot on ORG level to the master depot of this "specific" site.
  • Push mechanism: The Distribution Service is installed on the source and copies files from the source to the target. Several copy processes may be required because there may be several targets for each source.
    Example: A Distribution Service is installed on ORG level; it copies the files from the central depot on ORG level to the assigned master depots of the sites.

We recommend employing the Pull mechanism for distribution between sites. However, this calls for an individual Distribution Service in a site (which is responsible for distributing data in this site). Since you can only have one Distribution Service per site, the data is always distributed ("pushed") to the local depots within the site with the Push mechanism.

Example:
Region 1 contains Site 1 and Site 2. The master depot of Region 1 is also master depot of Site 1. Site 1 and Site 2 are connected via a slow WAN connection. The Distribution Service responsible for Site 2 resides on the Management Point within Site 2.

 

Thus, the distribution service of Site 2 runs only one of the four distribution processes via WAN:

  1. Reading the data from the source, which is the master depot of Site 1 (via WAN)
  2. Writing data to the master depot of Site 2 (locally via LAN)
  3. Reading the data from the master depot of Site 2 (locally via LAN)
  4. Writing data to the particular local depot (locally via LAN)