So you want to host your own Groove Servers

Many customers express concern about other parties managing Groove data stored on the Relays or Manager, and so they decide that they want to host their own Groove server infrastructure. Here are some important points to keep in mind while planning a Groove Server deployment.

·         Don’t make the mistake of planning to install all of the Groove Servers on the same 64 bit machine. While it might look good on paper, this is a bad idea. The installation may allow you to continue to install, but the last server installed will basically stop the previously installed servers from functioning. Also, some of the servers use the same port for different functions – for example Manager uses port 80 for administration, while Relay will listen for clients on port 80. While the installs might look like they will work, the previously installed server will not function at all … and no amount of tweaking will fix it

·         Similar advice applies to installing Manager with Auditing on the same machine. This configuration will work, with careful planning (see 927086 “Some manager functions fail if the manager service and the audit service are installed on the same computer in Groove Server 2007 Manager” at https://support.microsoft.com/default.aspx?scid=kb;EN-US;927086 for details), but in the long run, the configuration will cause some performance issues further down the road … particularly if you are very thorough when it comes to auditing what users do in Groove.

·         Virtual machines – many people have asked what our support policy on installing servers on virtual machines is. Officially, we only support the forthcoming Microsoft Virtual Machine environment. However, the servers will install on other virtual machine systems that support a 64-bit Windows 2003 environment. The catch with any virtual machine environment is that while it may work for a small-scale test scenario, some of our servers (Relay in particular) are resource-intensive. Consider typical Groove Relay operation. Because the clients send all data to the relay when they cannot communicate with each other peer-to-peer, relays tend to have high resource usage. Clients will send data to the relay and receive data from it, often as fast as they can connect, and the client will keep connections like that maintained, so the Relay needs to sustain many relay connections and write data to the disk and read it from the disk frequently. In a virtual environment, this “server” will be sharing those resources with other “servers” in that environment, so the user experience is most definitely going to be impacted by the resources other images are using at the time. Keep this in mind while planning - for example, don’t put a Groove Relay on the same Virtual Server Host that a heavily used SQL or Exchange server image is running on as well, because they need the same resources.

(Contributed by guest author Patrick Gan)