FOG implementation / storage node questions

  • Hi fellow IT Folks,

    First off, I just want to say that FOG is the best piece of software I have ever had the privilege of using, bar none. I like its flexibility, I like what it can do, and I like what I can do with it. I used to use DRBL back in the day, and once I got the feel for using FOG, FOG became the replacement, and I haven’t looked back since. So I want to give a wholehearted thanks to everyone who works endlessly to keep this project alive. FOG simply is awesome!

    I have a question that I am hoping someone can help me with. I work in a community college as a computer lab manager. Part of the way I have implemented FOG is to have a master FOG server on a cart with 2 to 3 other FOG servers serving as storage nodes. I then take this cart, plug all FOG computers into a gigabit switch, and then connect all of the clients that I want to image to the switch as well. At the most, there are 16 clients; at the least, there are 10. Taking the scenario where I image 16 computers at a time, I access the FOG web console and configure the storage nodes in the following manner:

    Master FOG server, 4 max clients assigned to this storage node.
    Storage Node 1, 4 max clients assigned to this storage node.
    Storage Node 2, 4 max clients assigned to this storage node.
    Storage Node 3, 4 max clients assigned to this storage node.

    This way, the imaging load is evenly distributed across the 4 nodes. (Or so I think.)

    When I image, I create a group in FOG that I put the hosts in, configure the image and OS associations and do a unicast restore on the group because if I try multicasting, it chokes the switch. Overall it takes about 30 – 45 minutes to push the image out to all 16 clients. The image that I am pushing out is about 40 GB uncompressed.

    Since I just took this job in August, I did a quick-assemble on the fog carts with what I had at the time – retired Dell Optiplex computers with 2 GB of RAM. I put 300 GB HDDs in all units, and I am only pushing the one image.

    I would like to better implement FOG in my next round of imaging while still retaining the carts and isolated server setup. Does it sound like I am on the right track? Do I have the storage nodes assigned properly for even load balancing, or is there a better way to assign hosts to storage nodes?

    I’m interested to know, how do you have FOG implemented?

    Thank you for your time, and I welcome any and all suggestions.