Jun 21, 2015

Which option is better, to have diskless server and boot it from SAN or servers with disk and allow them to boot locally from HDD?

We are planning for a hardware refresh. During discussions with h/w vendor, they suggested to have diskless blades and boot it from attached SAN.! I am not sure, whether this is good option in a long run.? I always thought, the local harddisk in the server was better option, as replacement is easy. However, the vendor informed we would be wasting hard disk space, by having large hard disk and build RAID 5 in it.
The best cost per value is with http://Jentu-networks.net, they provide a non-persistent golden image with 0.5 usage of the streaming server CPU and no spindle from the self healing drive array
Adding to what Number 6 posted, the 2nd part of the question is a little confusing to me. It runs contradictory to the previous query. First you are talking about having SAN - RAID, then about building RAID on a local drive (you would have to have at least 2 local drives to do this) But, ignoring that if you run RAID 5, you reduce total capacity of the drives in the array - it's the price of redundancy. The beauty is that you can pull a drive, put a blank one in its place and the missing data can be rebuilt from data scattered across the remaining drives. Spinning disk is cheap right now, so the loss is acceptable. Here's a cool RAID calculator that illustrates the cost of various RAID configs in terms of net drive space available.
Mean Time Between Failure. All drives fail, it's simply a matter of when.
Blade servers should boot from NAS if possible. Local hard drives work just fine, but spinning drives have the ugly MTBF problem. If a disk goes out in a RAID array, you pull it out and put in a new one. The array rebuilds itself and usually the only impact is temporary slow response time because the array is busy fixing itself. No outage. Local hard drives fail and the server fails hard.
Answer this