Date: Fri, 11 Aug 1995 09:59:00 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: msmith@atrad.adelaide.edu.au (Michael Smith) Cc: msmith@atrad.adelaide.edu.au, john@zyqad.co.uk, freebsd-hardware@freebsd.org Subject: Re: Upgrade to my machine Message-ID: <199508111659.JAA04184@gndrsh.aac.dev.com> In-Reply-To: <199508110937.TAA03196@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Aug 11, 95 07:07:12 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Rodney W. Grimes stands accused of saying: > > > Generally, the 1G disk will perform better than a 1/2 price 500M unit, > > > so two disks is false economy. > > > > This is absolutly the opposite of the real situation. 2 disk drives of > > 1/2 the size and identical performance characteristecs give you 2 spindles > > But that's what I'm getting at; if you take a 1G disk that costs twice as > much as some other 500M disk, it'll generally have _better_ performance > characteristics. And my point is that this is not ``generally true'', the performance delta in 500MB vs 1G drives is zero today. If you know what models of drives to look at and do some carefull agressive shopping: XX. BAS DEC3053L Dec/Quantum 535MB 3.5"x1", SCSI-II, 5400 RPM, 9.5mS $ 195.00 XX. MER FUJ-1606S Fujitsu M1606S 1.0GB 3.5"x1" 5400RPM 10mS $ 427.00 Those are my current best price point drives in both sizes. Note that my best 535MB drive is actually less than 1/2 the price of my best 1G drive, and it has a _faster_ seek time, and all my iozone and bonnie data says the DEC3053L is a clear winner over the 1606S. > > that can be doing data trasfer at the same time and with proper load > > balancing gives 2 times the over all performance. I have sold off _all_ > > of my 1 and 2G drives and now stack 535MB 5400RPM 4.4MB/sec drives up to meet > > what ever capacity I need. My make world times are down 45 minutes or so due > > It's nice if you can afford it; here the price point is on the 1G disks, > so that's what I recommend. It's an interesting approach, though. I'm > speccing up a new workstation, so I might see how it pans out as compared > to a 4G Micropolis M3243. The curve on $/MB falls off rapidly above 2G, but up to that point I can still stack these DEC 535MB drives up at a lower cost. At 3G the curve swings the $ point the other way, but 8 of these to make a 4G drive would be at $1560, making the M3243 a clear winner in the <$1400 range. But, performance would be _far_ better with the 8 drives _if_ you can load balance the application accross multiple spindls _or_ you had stripping technology in the OS you where running. > > > You can only boot from the first two disks in the system. IDE disks count > > > first, then SCSI, so you can only boot from a SCSI disk if there's only > > > one IDE. > > > > If you have only scsi disks you can boot from drive 5 if you like, unless > > someone again has broken that piece of code :-(. > > That's dependant on BIOS support for more than two disk drives; certainly for > the Ultrastor controller that's optional (and has caused problems with > other non-operating systems). I don't recall what the situation is with > other controllers 8/. For any current production performance class scsi controller (bt, aha, ncr) the above mutliple drive support in the BIOS is a given. Your talking about abosolete, no longer avaliable, low performance controllers if they are missing this feature. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Reliable computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508111659.JAA04184>