Maxtor announces ATA133

[otheos]

If you haven't seen it at my site (http://www.otheos.clara.net/) or anywhere else, Maxtor (or rather Quantum) made one more step.
It was kind of expected by many while others thought ATA100 would be the last parallel ATA interface. To be honest it's a marketing move (note here by Quantum's labs, now part of Maxtor) since ATA133 will not add any more to the already far from saturated ATA100. With current PCI design being limited to 133MB/s that practically gives aroung 110MB/s (depending on the chipset), the bandwidth increase from ATA100 to ATA133 will be marginal, and certainly unnoticeable
The new interface however brings two new features that will make it worthwhile as an update rather than an upgrade. The first is CRC error checking (like SCSI) which adds to the data integrity and makes IDE storage safer. The second and more important is the break of the 137GB limit on disks size by current ATA interfaces.
There you go then, expect it to hit the market this time next year, but with Intel not embracing the ATA133 interface, expect a direct jump to SATA "inside" your Intel PC in 2003.
Read the full press release here

otheos's site (http://www.otheos.clara.net)

[otheos]

Has IDE come to its end RPM wise? I see burst speed rates get higher, but not the RPM. IDE's been @ 7,200rp, for some time now...

http://ndc.digitalmntsnow.com/BBS/NDC_Mark.gif

[otheos]

Well 10K IDE drives will eventually come. But not before the complete SCSI line up makes the transition to 15K or faster like the 7200rpm SCSI moved to 10K last year. Remember, manufacturers do not want competition to their SCSI drives from IDE! SCSI has 10x the profit margins.

With the high platter density of IDE drives however, the benefits of high rpm is not as evident as in SCSI. Seek times the measure of a HD's responsiveness will obviously go down, but nowhere near current 10K (~4.5ms) SCSI drives levels.

Still it will be a nice improvement.

otheos's site (http://www.otheos.clara.net)

[otheos]

I also found out that Seagate will not be supporting ATA133 together with intel.

Promise will be announcing an ATA133 controller soon, to join VIA and ALI to the adoption of the new interface.

BTW, remember that MAXTOR needs the new intreface to push their >137GB hard drives.

otheos's site (http://www.otheos.clara.net)

[otheos]

Remember, manufacturers do not want competition to their SCSI drives from IDE! SCSI has 10x the profit margins.

Good marketing point there. But I have a feeling it has to with more than that... Perhaps IDE controllers aren't sophisticated enough to run data @ 10,000rpm?

http://digitalmntsnow.com/NDC/BBS/NDC_Mark.gif

[otheos]

rpms are not related to interface (at least not immediately). The ability to place the heads with precission over the data at such high rpm combined with the already high platter density will not prove a good idea.
Unless they use highest quality actuators the seek times will not improve at all (other than the reduced latency due to the higher rpm, the actuall time to find the data might even go higher). And this will not prove good as seek times are the most important feature of the drive in terms of performance.
Using high quality actuators will get the costs higher and with the small margins it's hardly a good idea to go out with a $300 80GB 10K IDE drive when the 7.2K drive is still at $150! On the other hand an 80GB drive at 10K and $300 with good performace will temp many SCSI customers.

As for the interface, the main dissadvantage of IDE is its FIFO characteristic (i.e. data is served with the order it was asked) and inability to disconnect. Having the higher STR that the 10K will provide will not impose any limitations are it is not expected to exceed 70MB/s in the near future (i.e. before S-ATA comes) and that's low enough for most ATA100 controllers (depending on driver maturity).


otheos's site (http://www.otheos.clara.net)

[otheos]

As for the interface, the main dissadvantage of IDE is its FIFO characteristic (i.e. data is served with the order it was asked) and inability to disconnect.

I wonder if it's possible to use "Command-tag queuing" instead of using FIFO for EIDE drives which handles processes more logically...

http://digitalmntsnow.com/NDC/BBS/NDC_Mark.gif

[otheos]

What for? SCSI is there to use if you need it. I doubt S-ATA has tag-queuing, but expect S-SCSI to pick up where U320 will leave it.

otheos's site (http://www.otheos.clara.net)

[Axel]

several questions -

why would you want to risk losing that much data when the drive eventually fails? Disk arrays are so much safer not to mention cheaper and in most cases faster than a single drive. 137GB - what's the point? You're talking server class machines here and they, for the most part, use disk arrays because it is too risky to have only one huge drive.

what are you going to back this up to? Are you willing to buy a case of DVD's and a $1500 burner to back up that hard drive? Are you going to buy a RAID card and another drive and mirror it?

I've got trouble trusting a 40 gb drive and, if you maintain a clean system - why do you need that much space? I've only just used up a little more than 10GB and that's because some of my games take up 4 and 5 GB alone if you do a "complete install" Some games are coming on DVD - so you've hit the 4.7GB space requirement if the whole DVD is full -

Here's what I see happening with that much space - people will keep loading and loading until the system crashes because of software conflicts - and then they end up reloading after some time consuming and often expensive support. I think it's healthy to limit the available space so people have to uninstall unused programs.

I know - I know - Joe. Q. User doesn't do that...... they just delete without uninstalling and cause themselves problems....

Another argument is the time it will take to defrag the disk -

Lastly - Maxtor recently announced financial troubles and losses, some coming from the absorption of the Quantum hard drive division. Their 3rd quarter isn't looking so hot..... and now the industry isn't embracing the new standard of 133....

My predictions -
The industry probably will eventually embrace the ata133 standard some time early next year and start building for it. Most new speeds see this until other manufacturers catch up and more people request it..... I think the only idea I haven't seen this on, sadly, is firewire..... One kiss of death for new technology in the short run seems to be - does Apple use it?

With the size of popular programs growing, we will need larger disk drives, but the speed of the size increases are far ahead of the need - probably by some 2 years.

I expect some of these larger disk drives to go to the 5.25 bay format again simply so it's cheaper to make the components. A drive with wider spacing of the components can release heat better and be less read-write error prone than a tightly packed multiple platter array.

[otheos]

Some good points there but not all drives are used for static.

You need a large ~100GB drive for various reasons and while a RAID array will make up for space and top the speed, you can still use the space and build bigger arrays.

I personally fill a 40GB disk in less than 35minutes from data crunching. 10 minutes later the drive is empy and a new session runs. 100GB? I'd kill for one right now.

Don't always take Joe for example and RAID5 is not always the solution for price/MB.

The 180GB baracuda for intsance stands at double the price of a RAID5 ide/scsi array of 520GB however it may well be preferred when there's no need for 520MB and can't justify the cost of the array.

Now backup is a strong point, and with current medium (corporate) backup systems in the 100s of GB's backing up this monsther is still expensive (and time consuming). But like I said, not all data is static, and there many many uses where data doesn't stay on the drives for long. A >100GB drive is usefull as your runs are no longer limited to 40GB.

As for Joe, ATA133 means faster than ATA100 and with the later being around for a while, ATA133 will sell (some 4200rpm junk with 12ms access time that is ...hey! 133MB/s fast!).

With serial ATA (intel backs it up) around the corner ATA as we know it won't be around this time next year.

What I always say, is given the price of SCSI, what a pitty it is seing IDE prevailing. IDE is less efficient, uses more resources, is not as flexible or expandable. Given that the price difference between SCSI and IDE is due to the sales volume only! and market sizes (i.e SCSI would be IDE priced too if it was mainstream) and hence artificial, isn't it a shame that SCSI is actually being dumped for an inferior interface? But that's me moaning....

Yes ATA133 is useless. For maxtor it means sales to people who like 10 more that 5 and 20 more than 10, and as a collateral, some low budget space starved low budget users will get something out of it. But that's as far as it goes.

[Axel]

I agree about the SCSI format - the server market is still strong with it comes to SCSI - just not the PC's....

one major factor - it's a matter of marketing - the same happened to the betamax videotape system - it was better and the tapes were smaller and a little cheaper to produce - it just wasn't marketed well enough.

One of the huge benefits to SCSI was multiple drives on the same ribbon - not just 2, but in many cases 5 or 7.

I think another thing that hinder's the success of SCSI in the PC is it isn't as simple to set up - sometimes you have to termnate the devices differently, you have to assign a device number - and most people don't take the trouble to understand that. Re - the success of PnP..... SCSI definitely isn't PnP right out of the box.....

If SCSI was as simple as a USB connection, then volume would go up and price would go down in a matter of months. By Christmas, 2002 the price would probably be real close to IDE/ATA technology.

As for the size factor - I'd argue that you and I might have "power user" needs when it comes to drive space, but most people in the next 10 years will not and won't take the trouble to learn enough about their hardware to get to that level. I base this on the "stupid user tricks" I clean up here in the office. If it weren't for bad programing and *.tmp files, they wouldn't need half the space they have now. The only thing that rivals tmp files for space on most office PC's are collections of e-mails the user should have deleted a long time ago or never gotten in the first place.

What do you think?

[otheos]

I think another thing that hinder's the success of SCSI in the PC is
it isn't as simple to set up - sometimes you have to termnate the devices
differently, you have to assign a device number - and most people don't take
the trouble to understand that. Re - the success of PnP..... SCSI definitely
isn't PnP right out of the box....

Now here I disagree: is it too hard to set an ID number (like you set master
slave) and terminate the last device?

Or is it easier to know the complete IDE white paper by heart to sort out
your CDRW? Set it as master on the secondary but wait, you can't use your
slave CDROM for on the fly cause target and source drives can't be on the
same bus, so put your hard drive there, but ohoh, you can't cause the CDRW is
PIO and the HD is ATA100 and I don't want to, but if I put my HD wiht my
CDROM and do some gaming it wont be able to swap and read from the CDROM so
I'd rather put the cdrom master sec and the burner slave but again I get
buffer underruns and with BURNPROOF kick in I get 12mins for a disc from my
super duper 24x burner, and erm, hold on what about that Zip drive that
doesn't always read the zip disks and only works with special IDE drivers
that oh no! make my HD slower and did I see a BSOD while ripping that CDDA
and I can't enable DMA cause the DVD sees movie discs as audio discs and it
resets to PIO after the reboot, and my motherboard doesn't like that master
is in the middle of the cable and slave at the end but my slave is further
away and my full tower is higher than the longest cable and I get CRC errors
if I use longer (out of spec) cables and the system freezes if I try to copy
and defrag, and after it reboots it will never finish the defrag because the
CDRom is not found if there's a disc in.......................

I guess that's easier. lol

I'm affraid that as long as Joe dictated Marketing and marketing dictates
R&D, we will keep taking technology advances as collateral audience.

Joe wanted IDE, so I can't buy a SCSI CDRW higher than 12x anymore! How do I
feel about it? I will have to pay more to buy a firewire drive then, and what
have changed for me? Sure, IDE works, but SCSI works better. And while >12X
worked in SCSI, IDE needs firmware tricks (Burnproof etc) to keep it going,
and I am asked to pay that as well!

So everytime you hear technology is cheap, just think of me giving a big
sigh! :(

[Axel]

I don't disagree with anything you've said, but how many Joe Q. Publics spend that much time with adding in their own CDRW's and ZIP drives. - They buy stock off the shelf and rarely ever have such conflicts because OEM's don't typically want to build systems that might have conflicts like that......and generate support calls.

How many cow boxes have you seen with a CD-Rom and CDRW drive in them? - what you see is an "upgrade" to a CDRW drive - but that means Gateway will totally replace the CD-Rom with the CDRW - no conflict.

so the OEM's don't do it - so all the problems you mention don't typically come up for Joe Q. P........

And now they have a DVD-ROM/CDRW combo out there to even remove that little conflict -

Most of the market is OEM - the cheaper $$$ gets the business. I think Packard Bell took it down to ruin for them - their last products were typically door-stops coming out of the box that were basically dumped on the mass-market..... I fear DELL is heading that way now.

So if the superior technology doesn't get the marketing to support it's launch and continued sale - it loses out to what the middle-class American will pay for. And we're headed into a ression.

Now - if SCSI had the marketing New Kids on the Block or Teenage Ninga Turtles got - then IDE would be a memory.....

Have I mentioned - I like discussing this with you - it's fun!

[flinder]

You Guys are great.
Thanks for your time.

[otheos]

Lol, it is fun isn't it?

Someone may thing that I really hate Joe, but it's Joe's buying power that makes technology advance and prices to be reasonable, but as I said earlier, not all is good and fair. But it never is!

[SalaTar]

Want to heat it up>?
SCSI is hanging on as the server world wants to pay for it.(we are a hard group to pass new info to as we accept mainstream as the only stream)
Things are not going ahead as HHD Company's wont embrace alternatives...
Mem, has finally gone to the point of zero....we should have a alternitive to the "gear factory" of old drives...


I will get a OUCH here:)

[otheos]

Salatar,

interesting Avatar. Does it mean FreeBSD is behind WinXP?

[SalaTar]

If freebsd was behind XP ,Xp might work..lol

No the avitar is my change in mood from Winblows to Opensource...
I Have been learning for my MCSE and after I previewed XP I changed my mind and have now decided that I want to use something else...