We have all heard that we need to split our data files from log files and even put tempdb on it's own disk(lun) as well. I had the misfortune of sitting through a presentation from a Netapp vendor talking about thin provisioning, using large storage pools, etc. Later I asked my storage guy if we use large pools of disk or if my luns were different arrays of disk. I was told that my luns are just carved out pieces of a large array. It was explained that we have a large set of disk that are formatted and I just have a slice of the disk, so my multiple luns that I thought were different physical disk all really come from the same spindles. Should I be alarmed?
Yes and No, It depends. Ok those will be your top two answers. Seriously though, with the density of today's disk and the cost of tier 1 storage this is the norm. What you really need to be concerned with is throughput and IOPS. I wrote a blog about it [here](
http://www.timradney.com/iops). The key is having the rights amount of throughput for your database. Having a lot of spindles being able to server up your storage as well as the storage of other systems should be a win for everyone. Otherwise to give you your 1TB of storage, your enterprise would have to eat the cost of many TB in order to give you the IOPS and throughput you need. Your storage engineer being able to still isolate your random IO from sequential IO will help them with handing out the right amount of cache for the need. It also gives them the ability to move around hot spots within the disk. If the IO of your data lun becomes problematic for the storage pool it is on, your storage person can move it to a less utilized storage pool. Is it a bad thing no, is it most cost effective yes. Is it always the right solution, no. What it all boils down to is money, otherwise we would all have large pools of SSD's. :)