I was just wondering what configurations and the reason behind them to all you fellow admins out there. I'm just looking for mainly how others setup the storage array on different servers. This might come in handy to others reading this thread as well. Well I will start with how I configure a DC: 6HDD's 2 each in a RAID1 (Mirror) I then create a single partition on each RAID array as follows SYS, DATA and LOG partition. SYS holds the OS and swap file, DATA holds the SYSVOL and AD Database, and last LOG holds the AD logs. This is done becuase each one has there unique access to the drive like LOG would tend to be written to more than read and DATA would be read more than written to and SYS would probably be read more than written to as well. For an SQL server I configure as such: 6HDD's 2 RAID1 and 4 RAID5 with a SYS, APP and DATA partition SYS and APP would both be on the RAID1 array and contain the OS on 1 partition and the remainder of the drive to the APP partition where SQL would be installed and then the databases being installed to the DATA partition on the DATA RAID5 array this is done so i can loose a single drive in the array without loosing the data is faster than mirroring and safer than striping as the loss of one drive you lose everything. For Exchange I configure as such: 6HDD's 2 in RAID1, the next 2 in RAID1 and the last 2 in RAID0 I create a SYS, APP, DATA and LOG partition I create the SYS and APP partition on the first RAID1 array and install the OS on SYS and Exchange on APP, then i move the log files to the second RAID1 array where I have created a LOG partition and then the mailbox and public stores to the RAID0 array in the DATA partition I choose a RAID0 array as the logs are mirrored in an event i loose the stripe and have to rebuild from the logs and it was also chosen do to it's speed over RAID1 where I have a mix and read/write going on on the RAID0 array. So how do you setup things and why?
Unix. RAID5 for everything if possible. Otherwise always RAID1 for the OS etc. and RAID5 for data. /var in it's own partition so for logs it doesn't decide to fill up and try to take over other partitions. /home usually on another partition files. Also can move the MySQL databases off /var into say /home somewhere. That way the logs won't effect the database if it fills up. Although logs are usually rotated through cron rotating anyway. Sorry, but I refuse to use Windows servers =) Unless I absolutely have to, and that's only if a program really needs to run on it, such as AutoCads licensing manager. Otherwise it's Unix/Linux all the way baby. I have a hugely different philosophy than most Admins and Consultants. Most will use Windows, and it's good because there's a lot of maintenance, patches etc to keep up with, so they make a bit more $$ out of it with support costs etc. Me on the other hand, I like to use Unix/Linux and just have the things constantly running, very very minimal downtime, good redundancy and plans if things do fail, how do I get a replacement in within no time. Sure I don't make more money out of it, but at least the network continues to run easily if I go on Holidays to say the Bahamas.
