• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

SCSI now or serial ATA later?

mcveigh

Diamond Member
Thinking about going SCSI but was wondering if I should wait for serial ATA? If SCSI what would you reccomend? (cheap and linux comapatible)
 
Do you have a specific type of application in mind or is this just for general use? And what's your budget exactly? Hard to give a good suggestion without some info...
 
just general use but I do want to turn my primary amchine into a web and mail server.
I'd like to go scsi for less than $400
 

I don't think there's a such thing as "cheap" scsi , though with 400 $ US , you can easily afford at least one hard drive with a decent adapter.

When you say you wanna go " Scsi " , what do you want ? Cd-Rom , CD-RW ? or just a HD ?
 

If you want to save a bit of money you might want to look into the Tekram scsi adapters , cheaper than adaptec adapters , and you might wanna look into the Maxtor/Quantum Atlas line of scsi HDD's.

A large scsi HD will run you a lot of $$ though .
 
How expensive will it serial ata be in respect to parallel ata and scsi? Also you said it will be a year off or so, do you know of a more specific time frame?
 
From the manufacturer comments that I've read and industry review from IDF and other forums and conventions it sounds like SerialATA is going to carry less then a 10% price premium over ParallelATA drives/controllers when released. As for timelines; well Fujitsu, Quantum, Seagate, and WD have all demoed drives at Comdex and/or IDF and a few manufacturers have demoed controller cards, and I think even one Manuf. has demoed a SATA integrated Motherboard chipset, so I'd say expect to start seeing controllers and drives late summer/fall. (I haven't seen any published firm dates from manufacturers yet ... sorry, but trust that I'll post here when I hear something)

Keep in mind the final spec was released last August.

Thorin
 
If you're debating SCSI vs ATA, then SerialATA shouldn't factor into the decision, because when it is released, nothing will be different. When SerialATA is released, all the advantages that SCSI has, it will still have, all the advantages that ATA has it will still have, with the one exception to those being the cable length. SerialATA allows for longer ATA cables and they are smaller. From a performance aspect, SCSI still maintains all its advantages, since it's the drives that dictate performance, not the interface. Of course SCSI also maintains the bloated coast associated with that performance.
 
pariah
What are you talking about. SerialATA has the exact same interfce advantages as SCSI in that all the drives can 'talk' at the same time and don't have to wait on one another as they do with ParallelATA. The only difference is that you can't have 8 or 16 drives with SATA as you can with SCSI (although I suppose you could with enough controllers). I agree that when SerialATA is released the drives will not be pushing 150mb/sec although the bandwidth will be available to them. However by the time they are released they'll likely be up to 50MB/Sec (sustained) and more then one will be able to transfer that much data at the same time. (This is a huge benefit over current ATA). Current ATA drives will obviously not be made faster simply because of the new interface however they won't have to wait on each other. (I'm not saying SATA is a SCSI killer but it brings ATA drives alot closer to SCSI performance levels and is not going to be anywhere near the price of SCSI).

"From a performance aspect, SCSI still maintains all its advantages, since it's the drives that dictate performance, not the interface."

Ummmm so you think SATA is holding the drives back somehow? Please explain, that doesn't make any sense.

Thorin
 
FWIW, a Seagate rep. told me that a SATA drive would be on the market in March, this year, the same drive as what was demo'ed at Comdex.
 
When SATA comes out, the drives will still be the limiting factor. I don't think the drive manufacturers will cut their own throats by introducing 10K and 15K SATA drives while they can reap the margins from SCSI.
 


<< "From a performance aspect, SCSI still maintains all its advantages, since it's the drives that dictate performance, not the interface."

Ummmm so you think SATA is holding the drives back somehow? Please explain, that doesn't make any sense.
>>



You won't be seeing any 15,000 rpm S-ATA drives any time soon.
 
"SerialATA has the exact same interfce advantages as SCSI in that all the drives can 'talk' at the same time and don't have to wait on one another as they do with ParallelATA."

So what? That will yield about 0% performance benefit to 99% of IDE users out there. IDE already has rudimentary ablilities to do that and it has been shown time and time again, that 2 drives on one channel are not as crippled as some woul have you believe. Even so, most people with 2 drives put them on seperate channels eliminating the problem altogether. How often do you think people have 3 IDE drives or more running all out at once for long periods of time? This situation is so rare that ATA adding this feature without allowing more drives per controller is basically pointless. If you need that many drives running at once, chances are, you're already using SCSI. This feature is overrated performance wise.

"However by the time they are released they'll likely be up to 50MB/Sec (sustained) and more then one will be able to transfer that much data at the same time."

Where have you been? IDE drives already do hit 50MB/s, and putting them on SATA won't make them any faster, even 2 of them.

"Ummmm so you think SATA is holding the drives back somehow?"

No, the exact opposite. Parallel ATA drives are not being held back at all by the interface, so giving them a faster interface isn't going to do anything. Putting a 7200RPM IDE drive on SATA will not make it magically perform like a Cheetah X15-36LP. In 99.999% of situations for home users, the performance of drive dictates storage performance, not the interface.
 
Anyone know if tekram controllers are supported for Linux? I remember hearing good things about them at storagereview.com🙁

I think newegg has new quatum 10k rpm 18gb new for like $180.00
 
They all should be, if not through Tekram, then through 3rd parties. The U160 are based on LSI chipsets, so any OS driver that you can find on their site will work with the card, which include Linux and just about any other OS you might use.

I don't think Newegg carries Maxtor SCSI drives, at least I couldn't find them.
 
"IDE drives already do hit 50MB/s, and putting them on SATA won't make them any faster, even 2 of them."

Hmmm last time I checked storagereview.com the top spot was about 40MB/Sec or 42MB/Sec. Please show me a review of a ATA drive that does 50MB/sec sustained. And as I said and I quote "Current ATA drives will obviously not be made faster simply because of the new interface however..." I never said putting a drive on a new interface would make the drive faster. I said it would allow you to transfer across multiple drives faster.

"Putting a 7200RPM IDE drive on SATA will not make it magically perform like a Cheetah X15-36LP."

Again as I said and quote "I'm not saying SATA is a SCSI killer but it brings ATA drives alot closer to SCSI performance levels and is not going to be anywhere near the price of SCSI".

"In 99.999% of situations for home users, the performance of drive dictates storage performance, not the interface."

So you think average home users run web and mail servers? I didn't know that. (See the poster's second message in the thread).

It's this simple: Waiting for SerialATA will give you a performance boost, although perhaps not as large as going SCSI. Waiting for SATA will allow you to use your current drives (not having to spend lots of $$$ on scsi drives). If you're going to replace your drives anyway by the time SATA is here much better ATA drives will exist (this fall). Waiting for and using SATA will cost alot less then SCSI (whether you replace everything or just get a new controller).

Thorin
 
The WD1200BB/JB hits 50MB/s. The IBM 120GXP isn't too far off itself at 48MB/s or so.

As for the rest of it, I'm not going to waste my time debating this with you anymore. For whatever reason you have a habit of drawing the exact opposite conclusion from what I am trying to convey. I thought it might be me being unclear, but 2 other people here seem to understand what I am saying.
It's all speculation at this point. When SATA does hit the market and it doesn't perform the way you seem to think it will, we will resume the debate then.
 
"The WD1200BB/JB hits 50MB/s. The IBM 120GXP isn't too far off itself at 48MB/s or so."

Yes it hits 49MB/Sec on it's outter edge but if you look at the average sustained transfer rate it's around 39MB/Sec. (49MB/Sec outter, 29.1MB/Sec inner).

Edit: Correction I was looking at the review for the plain BB model the BB-SE or BB-JB model scores 43.8MB/Sec outter, and 27.9MB/Sec inner (average sustained transfer rate = 35.8MB/Sec).

As I said I'm not expecting SATA to be a SCSI killer, although you seem to think that's what I expect. I simply expect it to bring a number of benefits that SCSI already enjoys, such as hot plugging/swapping, smaller cables, longer cables, lower electrical requirements, simultaneous drive access, etc.... Given drive improvements by release and interface benefits maybe 15 or 20% over PATA, that's all I'm saying. And I repeat for a 3rd or 4th time now, I'm not saying it'll be a SCSI killer.

"I thought it might be me being unclear, but 2 other people here seem to understand what I am saying."

Yes two other people agreed that HD Manuf's wouldn't release 10k or 15k SATA drives anytime soon. Which you didn't even mention. If you think I keep coming up with the "opposite conclusion from what I am trying to convey." Then you aren't being clear. <shrug>

On a side note. Mauf's are missing out on alot of money if they are simply not making 10k and 15k rpm IDE drives because they think it will eat in to their SCSI sales. As seen in this SR poll: "Would you buy an IDE drive that featured a 10,000 RPM spindle speed if it were less costly than a SCSI drive yet more expensive than a 7200 RPM IDE unit?"
No
Votes: 114 or 19%
Yes
Votes: 486 or 81%

Thorin
 
I understand what Pariah is saying just as clearly as I understand what you are saying Thorin.
And I do think you are misconstruing what he has posted.

FWIW I don't think this is done deliberately, you two are just not on the same wavelength with your opinions:

Thorin: S-ATA will provide a performance benefit when the new interface is used with newer (S-ATA designed) drives;
and can possibly have a side benefit for older drives that can be adapted to the interface.

Pariah: S-ATA (on paper) shows no signs of providing enough appreciable benefit for current (or older) IDE drives to make
it anymore a viable solution for the average user than simply staying with Parallel ATA. If the user is considering an upgrade
to S-ATA then they will need to do some additional analysis to determine the best setup that they can configure, and also
may have to weigh in other options (such as SCSI) in that analysis.

Both statements are true (IMHO), but you are talking from different sides of the fence.


 
I think I heard that we will see Serial ATA in Q3 or so. But you have to expect that drives and the interface will carry too high a price premium for about 6 months while production is put into effect...I think it will be an affordable option (price/performance-wise) in about next year.
 
Okaaay... kinda took off since this morning. mcveigh, Serial ATA should be of no importance to you. For so-called "general use" and web/mail, you're primarily concerned with read capabilities for small files. You don't need need throughput (unless this is going to a monster web server, or serve up a lot of video clips/huge graphics - which it doesn't sound like). What you need is fast access. About the only thing SATA offers besides easier configuration is better throughput, which current and near-future drives can't fill up anyway. Sure, there are RAID arrays which might get close, but RAID helps very little for small file reads with low throughput requirements - thus the typical association of small-computer RAID with video editing, where file sizes get very large. On the other hand, the better access times of SCSI drives will make a big difference to you. I see two good options for your budget -

1) The El Cheapo - get an IDE WD 100GB SE. Whatever they did to the buffer and firmware in this drive, they managed to work some magic. It won't perform as well as a good SCSI drive, but will be a lot cheaper at ~$250. If you need serious capacity for your server, this may well be the only viable solution for your budget.

2) The SCSI solution - look for an older Ultra2 (80MB/sec) controller. I was scoping out Pricewatch last night, and the Adaptec 2940U2W and the Tekram DC390U2 (?) are actually almost the same price, around $130. Both are supported in Linux. You could also shell out the extra $30 or so for a U160 controller, but you're not going to need that kind of throughput for your usage. Pair that with the Atlas 10K III - I believe the 18.4 GB was ~$200, 36 GB around $300 (?). That's a $330-$430 solution, plus cables, which would be very effective for you. If you don't need much capacity (maybe you have a spare big IDE drives to store your personal stuff on), you could spring for the 18 GB Cheetah X15 36LP for ~$300, assuming you don't mind having a hot and noisy jet engine in your case.

Hope this helps...
 
Don't waste your time waiting for Serial ATA. It's not going to offer the same advantages SCSI already has (for many years) enjoyed. Sure, you'll be able to use smaller cables. But is that really something to get excited about? 🙂

SCSI is where it's at.
 
Back
Top