2x SSD as RAID-0 in 2nd optical bay, orig. must function... also RAM ?

Question

Hey all I know this is kind of a long post but I think it would help a lot of people out to get all of this information together in one spot so please help me figure this out. Thanks in advance.
I am purchasing a 3.33GHz Hexcore MP, aside from processor stock from Apple. My plans are:
Add 2x 4GB RAM to 2 of the 3 orig. 1GB chips, for a theoretical 10GB (for now) Is this feasible/advantageous?
2x 50GB OWC Mercury Pro RE SSDs as RAID-0, boot volume, in 2nd optical bay. The catch is I'd really like to keep the std. optical drive functional in the first bay.
4x 1TB 7200 drives in HD bays, two RAID-0 arrays, second as mirror of first.
So here are my series of questions:
1 - How can I hook up the 2 SSDs in the second bay as far as SATA/power are concerned. I do not want to hook them up externally and want a good fast pipeline for them, yet I do not want to disconnect the stock optical drive. Do I need any additional hardware/cables for an internal connection? I've also heard of a sort of hardware RAID enclosure that houses the two SSDs in the optical drive, occupying only one SATA/power connection and sharing it between the two of them. Does anyone know any specific products, and is the performance as good as two dedicated SATA connections?
2 - using Disk Utility (or similar), can the second RAID-0 2TB HDD set mirror the first, like two RAID-0s nested in a fat RAID-1, or, in a pinch, is Time Machine or cloning the way to go for redundancy there? The idea here is to not buy the RAID card.
3 - Can I partition a RAID-0 stripe or is that just a bad idea? The idea is to partition off space on the two RAID-0 HDD sets to mirror the SSD set (thus cloning the boot drive), plus a small partition for a windows VM. Any better ideas how to accomplish redundancy for the boot volume and isolation for a Bootcamp/VM partition?
I know this is a lot and may seem rather amateurish, but I'd really appreciate some help here. I'd like to get this ball rolling! Thanks in advance for your replies!
Steve

Answer 1

What are some of the 'other things' you refer to in this area that would yield a greater payback?
• The biggest payback comes from having a separate System/Applications/Paging drive, by moving the Users folder off it completely.
When you are running a program, it will read a User file, execute a little, then move into an area of the program or System that is not currently resident in RAM memory. This requires an access to the System Drive, to get a part of the program, a part of the System, or a part of a Paging file. Each access to the System/Applications/Paging area or drive takes longer because the arm was positioned to read the User file, and has to seek a long way to get to the System etc. area.
Each time you access System stuff, this moves the arm away from your User input file, and the next User read will take longer because the drive arm has to be moved back to the User file.
Separating System/Applications/Paging onto a separate drive from all User files speeds up everything you do.
• If this drive is fast, that is better.
Access time on a spinning disk varies varies, but it is commonly 10 to 20 milliseconds. It consists of seek time (time to move the arm to the track desired and verify that it has arrived at the proper track) and rotational latency (average time for the platter to spin around so that the required block is under the read/write head).
At 7200 RPM, (120 spins/sec) one full spin takes 8.3 milliseconds, and half that amount is about 4.2 milliseconds. At 10,000 RPM, (166.6 spins/sec) one full spin takes only about 6 Milliseconds, and half that is only 3 milliseconds.
A drive with a faster seek time, a higher RPM rate (which indirectly gives a faster transfer time), and a larger cache (approximately in that order) will speed up your System/Applications/Paging drive and make your Mac seem faster. For example, using a small 10,000 RPM Velociraptor instead of a 1TB 7200 RPM WD Black is snappier overall. It seeks faster and is spinning faster so the arm is positioned faster, the data get into read position faster (on average) and once reading has begun, the data are presented to the drive cache faster.
Access time plus transfer time is usually dead time, in that little else can proceed until the data transfer is complete.
• If this drive is a \[no seek time, no rotational latency, fast transfer rate] SSD, that is Much better.
An SSD eliminate both seek time (positioning the arm to the track) and rotational latency (time waiting for the disc to spin around to the right block to be read) and substitutes a miniscule (by comparison) read time. Once reading has begun, a huge block of data is immediately available for transfer, so the effective data transfer rate is also much higher.
• If this drive is an SSD Striped RAID, that is marginally better still, because it increases the effective transfer time.
But the access time, the Big Boat-Anchor in this process, was almost eliminated by using the FIRST SSD. None of the typical seek-time and rotational latency gains of RAID-ing spinning disks are recovered by RAID-ing SSDs. The transfer rate is increased, often nearly doubled, but that is the smallest part of the delays caused by disk accessing.