[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: re: Memory, Stata 10, and 32-bit XP

From   Phil Schumm <>
Subject   Re: st: re: Memory, Stata 10, and 32-bit XP
Date   Fri, 22 Feb 2008 11:49:43 -0600

On Feb 22, 2008, at 11:01 AM, Fred Wolfe wrote:
Our group is switching over to the Power Mac for some of the reasons enumerated by Kit. We are wondering if Kit (and others) have experience with the extent to which performance is boosted (+ or -0) by Raid 0, 1 or 5) using Stata, and if there are any recommendations.

I assume you're not talking about a Power Mac (PowerPC-based macs are no longer made), but rather one of the current offerings (i.e., Mac Pro or Intel-based Xserve). At the risk of giving a naive answer, let me point out that RAID 0 (striped) can improve read and write performance, RAID 1 (mirrored) can provide limited increased read performance (but only if the OS supports it) at the cost of a small reduction in write performance, and RAID 5 (striped, parity distributed over multiple disks) provides improved read and write performance. RAID 1 and 5 are robust to the failure of one of the disks, while RAID 0 is not.

Note that software RAID is not the same as hardware RAID; the former is not as robust nor will you see as much of a performance increase. OS X is capable of RAID 0 or 1 by itself (i.e., software RAID), if you have the disks available. We use software RAID 1 (mirrored) on our Xserves for redundancy, and it has worked well for us in the past. However, it is not comparable to a true hardware RAID.

Since Stata does so much in memory, I think the impact of RAID on Stata use will depend pretty heavily on what you are doing. If you're doing a lot of reading and writing from disk, then you should expect to see some performance benefit *if* you have a RAID implementation that can provide it. If not, then you won't.

Perhaps you knew all this already. If so, I'm afraid I have no experience in running OS X with a true hardware RAID, much less in using Stata on such a machine.

-- Phil

* For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index