Dear All, getting a bit confused here with the block size when formatting VMFS.
So when you create the array or LUN on your SAN it will ask you the block size you wanna use for your partition. So lets say you choose 128K for that LUN.
Later if you format this partition in ESX into VMFS with block size of 1024, what happens to that partition?? 1024 or 128??
as duncan says, the block sizes are different, when creating the VMFS the block size you choose will determine the maximum size of any vmdk's you create
for instance if you choose 1mb block size the max size vmdk you could create is 256gb.
I guess I didn't put my question right. Put ESX aside, all filesystems have their default block sizes. NTFS, EXT2, EXT3, FAT32 ... Yes?? for example Fat32 has a max block size of 32 K.
So if I have a LUN with block size 512 K, and then I present thisLUN to a windows server, and format it into FAT32. Whats gonna happen to that partition from a block level structure point of view.
Will FAT32 OVERWRITE the existing data structure?? or what happens??
anybody??? or is it that my question is not making any sense??
The reason I'm asking is becasue I'm confused with the way it works.
On one hand it is recommended to have small segments (8k or 16k) for partitions for SQL database, but on the other hand it is also recommended to have large block sizes for VMFS.
So if I want to deploy a SQL server and want to store SQL databases in virtual disks, what should I do about the blcok size when I create the LUN??
VMFS block sizes don't translate to your guest OS block sizes. VMFS uses regular blocks, as well as sub-blocks when it retrieves information off of storage. That does not directly correllate to the guest. For a guest, you create a raw disk, that raw disk is spread over VMFS. Inside that raw disk, your guest OS block size is how the blocks are laid out for the OS. You can use 8k block size within the OS, and you won't be wasting space on the ESX side, because ESX will use sub-blocks to address files smaller than its filesystem block size. The only real reason to use larger block sizes is to create a larger guest os disk, or larger file in VMFS terms. It does not give noticeable I/O benefits from what I've seen.
-KjB