I want to set up a small Web server cluster with two servers and a SAN probably using iSCSI. Do I have to install separate versions of the OS for each connecting server that will boot from using the SAN? If yes, how does this work?
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
I don?t want to have to enter in configuration settings -- IIS settings, Cold Fusion settings, etc. more then once. I also don?t want to have to separately configure each machine when I install certain software -- obviously this would get much worse as I add in more then two machines.
I have never tried SAN boot using an iSCSI adapter. Usually, iSCSI adapters don't come with a boot
. Using a standard network adapter and an iSCSI driver would require the OS to be loaded for the iSCSI driver to work.
You can create a cluster with internal boot drives in each server then use
for the connection to external storage including the quorum disk for NT-based MSCS clusters. This is a good way to create what I call a "stretched cluster" where each node can be placed in different physical locations.
You mention IIS so I assume these are Windows servers. I would use FC adapters for booting into the SAN. You can use
Lesson 10 of SearchStorage.com's SAN School
for some tips on how to accomplish that task.
SAN boot requires each server to have its own exclusive copy of its operating system disk. You cannot share the C: drive between servers in a
. Windows is not VMS, which does allow a "shared root boot" disk but this should not be a problem for you. Using a MSCS cluster your IIS installation only happens once for the cluster. The cluster will then fail over the IIS service between nodes automatically. As long as your application software is cluster aware, the cluster looks like one logical instance of that software. All settings and state information is shared between cluster nodes through via the external quorum resource of the cluster. This allows automatic fail-over and fail-back of applications between nodes.
For applications that are not currently cluster aware, you can either write your own cluster resource DLL file for the cluster, or run the application under the "generic" cluster resource. Using the generic resource, you can even make Windows notepad sessions fail over between servers in the cluster.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our
The views and opinions expressed by Christopher L. Poelker are his alone and not necessarily shared by Hitachi Data Systems.