Appendix: Using Nodes

Deploying the Marratech Manager server requires planning, mainly because of the bandwidth loads it may create.

A typical situation is that of two offices wishing to communicate across their intranet. The intranet connection might not have enough bandwidth for multiple users connecting to a server located remotely. Every such connection represents a duplication of data.

Normal Connection

For example, the following picture illustrates three users needing to connect across the intranet. The Manager is installed at the office on the left.

All three user connections need to use the intranet connection to participate in the meeting. This is a waste of network resources as most of the traffic is duplicates of the same information. With many users, even fast intranets can be quickly overloaded.

 

Marratech solves this situation with the introduction of node functionality. A remote node is a subset of the main Manager (called central node). When a remote node is installed on a server located at a different location, it finds the central node and "connects" the traffic between them. This connection saves considerable amounts of bandwidth as no more duplicates are sent between the two network locations.

Using Nodes

As you can see in the illustration to the left, only one connection is made instead of three, thereby saving significant amount of bandwidth. Only one connection goes across the intranet. The remote node then takes care of distributing all the required data within the LAN (Local Area Network), where bandwidth is usually not an issue.

 

The Marratech client automatically finds the node that is the most appropriate when joining a meeting, removing the need to instruct users on where to connect depending on their location. This is done by Marratech first receiving a list of available nodes and then testing their accessibility, performance and if Multicast is available.

Several Nodes

Several remote nodes can connect to a central node in order to save bandwidth across multiple locations as illustrated here.

 

divider

Here are some more details:

  • A remote node may only connect to a central node
  • Up to five remote nodes may be deployed depending on your license.
  • A remote node may be deployed on a different operating system then the central node
  • Users must still have access to the central node for logging in and finding the appropriate meeting room.

divider

1. How to deploy a node - Central Node


To deploy a node, follow these steps:

  1. Install the central node (main Manager server) using administrative rights and configure it to your liking using the other chapters of this manual.

  2. Log in to the administration pages and click on the Advanced link.

  3. Under the Session Engines listing (where you will see the current central node), press the Add New Module button.

  4. Hosting Node will be of the type "New Node" automatically. Press Next.

  5. Give the node an alias that is easy to understand. For example, the city in which your remote office is located. Avoid accented characters and spaces.

  6. Choose between an "active" or "passive" connection model.
    • An active connection model will make the remote node look for and establish a connection to the central code. This is useful if your remote node is located behind a NAT firewall.

      NOTE: We strongly recommend creating port forwarding rules in the NAT firewall. Certain NATs have a tendency to switch ports after a period of time, which creates connection problems between the Central and remote Nodes. Possible side effects include empty e-meeting rooms and one way communication.

    • A passive connection model will make the central node look for and connect with the remote node. This is useful if your central node is behind a NAT firewall.

  7. In the following step, you must specify the IP address of your central node or remote node.

If you choose to create an "active" remote node:

Your remote node will automatically look up the central node. Normally, no modifications are required when choosing this option and you may simply press Next.

Central Node

Bind Address Specifies which network interface the central node should wait for connections on if you would like to specify what interface to use.
Bind Port This is the TCP port the central node awaits a connection for. The Bind Port is used for transmitting status information between the nodes.
Tunnel Port This is the actual meeting data (UDP) port used to receive and send audio, video, wb, chat, etc… between the nodes.


New Node

Connect Address The central node's IP address is used here by default. This is the address the remote node will attempt to connect to. If the central node is located behind a NAT firewall and port forwarding is used, enter the external address of the NAT firewall here.
Connect Port The remote node will connect to the central node on this TCP port.
Tunnel Port The remote node will use this UDP port to send and receive meeting data.

If you choose to create a "passive" remote node:

Your Central Node will initiate the connection towards the remote node. Normally, only the Connect Address needs to be filled in, since the central node can not guess where the remote node will be installed.

Central Node

Connect Address Empty by default. Please enter the remote node's IP address. This is the address the central node will attempt to connect to. (If any NAT is involved, you must enter the public address.)
Connect Port The central node will connect to the remote node on this TCP port.
Tunnel Port The central node will use this UDP port to send and receive meeting data.

New Node

Bind Address Change only if you have multiple network cards on the remote node. If so, specify which network interface the remote node should wait for connections on.
Bind Port This is the TCP port the remote node awaits a connection on. The Bind Port is used for transmitting status information between the nodes.
Tunnel Port This is the actual meeting data (UDP) port used to receive and send audio, video, wb, etc between the nodes. The remote node will wait for meeting media from the Central Node on this port.

 

divider
  1. >Now you may choose a startup procedure for the module. However, this is mainly for future use, just leave the default setting - Automatic.

  2. Creating a node will now take a few moments. Once your node has been created, you will need to restart the central node. This will close any active e-meetings that are currently ongoing.

  3. Once your Manager has been restarted, proceed back to the Advanced tab in the administration pages. Press the Download node start kit button next to your new node found in the Nodes section, and store the downloaded node.kit file

 

2. How to deploy a node - Remote Node

 

Now on to installing the software for the new remote node.

Install a Marratech Manager server software on the remote server, make sure to have administrative rights when doing the installation.

  • Do not change anything from the initial configuration.

  • Do not install your official Marratech license file.

  • Do not start the new installtion.

Once you have completed the install, do the following:

1. Open your Manager install directory.

Mac OS X Specific Hint

On Mac, the Manager is in a hidden directory called .MarratechManager.app (notice the dot in front of the folder's name).

To modify the node.kit file or to access the HTML / JSP pages easily from the Finder, simply create a shortcut by opening a Terminal and typing (or copy pasting) the following command in a single row.

sudo ln -s /Applications/MarratechManager3.5/.MarratechManager.app/Contents/Resources/Java/ MarratechManagerShortcut

This command will create a shortcut to the main folder within the hidden Marratech Manager folder. You will be able to see this folder by opening your Applications/MarratechManager3.5 in your Finder.

2. Copy the node.kit file you downloaded from your Central Node into this directory. (You will replace the present one)

3. Manually remove the file license.kit from the installation directory.

4. Start the remote node, either as an application, or as a service / daemon (recommended). (Please consult the start / stop chapter.) It will either attempt to connect to the Central Node (active) or await a connection (passive).

The remote node may now be fully administered from the central node via its web administration interface.

Hint Just as with a Central Node, a remote Node writes log info to an error.log and Manager.log files. These may contain valuable troubleshooting information.

Hint From the Advanced page, you are able to see the activity on each node, by looking at the Session Engines, their log files and restart / shutdown the various part of your deployed solution.




    forum    support Copyright © 1998-2006 Marratech AB