You're kinda going in two different directions here, technically, as well as ...say compliance sake. There isn't an easy black and white answer.
In general, how do 'YOU', define redundancy? I ask that not to be funny, but it does vary based on the engineer. In it's simplest form, I would recommend you configured the DAS as a RAID10, as we initially did. The caveat is, based on SLA's, and compliance that would not work, because all the drives were in the same chassis and essentially only, "logically" separated. As a temporary compromise, I setup up space on my SAN essentially setup a copy there. The client thought it was good.
2. Using the above as a catalyst, I took it a step farther, and created a TRUE redundancy. See, as I define redundancy, is similar to an HA cluster. One goes down the other peer component is there to take over. Thus, the above still wasn't redundant in the sense I speak of. We bought another ELM with another DAS50 and configured them as redundant. Thus if one goes down, logs may be redirected.
As far as best practices are concerned, I like to look at the combo's as an example.... The true storage ALWAYS needs to be separated from the main device, thus no single point of failure. Secondly, you can also attach to elms to the same DAS chassis, as the chassis has two separate controllers.
Hope this helps Kamal,
Thanks Joe.. I got you.. this really helps me..will connect with you if I need any help.