a) Neither RAID 0 nor RAID 5 improve access times very much, so they won't make booting or loading programs very much faster. The only way to improve access times is to use a disk which has a faster access time, such as a Velociraptor or an SSD (Solid State Disk).
b) RAID 5 has very poor write performance. Don't use it if you're depending on writes being fast.
c) For two disks in a RAID 0 set, you have TWICE as much risk of loosing ALL your data as you do with a single disk.
IMO, The catch-all performance and safety winner for home users is four drives in Raid-10.
You get improved read performance from the Raid-1 components but not improved writes. On top of that you get improved reads and improved writes from the Raid-0 component.
(Raid-10: A Raid-0 of two Raid-1 arrays.) (well, with four drives it is.)
From a reliability standpoint you are 100% protected from a single drive failing and 60% protected from two drives failing at the same time. (Statistically there are six ways that two of your four drives can fail, and Raid-10 survives four of those.)
Raid-5 works but has potential major flaws with large disks. Search for Raid-5 URE.
Basically the problem is that if one drive is failing, chances are high that another drive may also be failing because they have all been running the same amount of time in the same conditions.
Add to that the problem that if a raid-5 drive does fail, you have to put a new drive in to rebuild it. During that rebuild there is no extra redundancy, so if there is any sector anywhere on the remaining drives that can't be read, the rebuild process is guaranteed to run into it. If the sector can't be read at all the rebuild cannot complete and your data is lost.
Many I.T. types are saying that is a critical flaw in Raid-5 when using large disks, and have dumped the technology entirely.
I've been very happy with the performance of raid-10 in multiple situations.
1: Take the budget that you were going to spend on (I assume) 3 hard drives, buy slightly smaller drives and spend that budget on 4 drives instead.
2: Buy Western Digital Black drives. Google search for WD TLER and follow the instructions to enable the Time Limited Error Recovery bit on your new WD Black drives. (This is the reason I mentioned WD Black.) The WD Caviar Black drive is exactly the drive I recently used for an 8TB Linux network access video production server. They have been fantastic.
3: Alternatively, buy WD Raid Edition drives (RE2, RE3) which already have TLER turned on, but have slightly better logic for it and also do a little better job in RAID.
4: DO NOT attempt to use "Green" drives that lower their spin speeds when not in use. That behavior is asking for your RAID array to constantly lose drives and go into rebuild mode, or fail outright.
5: The reason I recommended WD and not other brands is the other brands do not have a TLER bit to flip, and do not make RAID edition drives. The TLER feature prevents drives from dropping out of your array when they encounter a recoverable bad sector. Consumer drives tend to spend a longer time recovering that sector, trigger a time-out on the controller, and get dropped.