Sign in to follow this  
MixedMedia

Seagate Savvio 2.5-inch, 10k Rpm Drive

Recommended Posts

Here is the Product Page. With a seek time of 4.1ms and 3ms latency, it sits between existing 10K and 15K drives.

Since it is limited to 2 platters, I expect 4/6 platter 3.5" 10K drives will be the cheapest ($/MB) and largest SCSI drives. When will be see 292MB drives?

Share this post


Link to post
Share on other sites

Given time, yes I think 2.5" drives (read: something smaller than 3.5") will become the norm.

The fact is that most people don't need anywhere near the drive size they have. I'm probably about mid-range on the SR scale of disk mania, and I don't need more than about 4-10G in most of my systems. Those of us who need lots of disk usually are better served by having it all in one place - lots of machines with 80G drives 10% full is a lot less useful than a server with .5T, and a bunch of 4G drives with OSes on them.

With no floppy, a slim CD/DVD, and a 2.5" HDD, SFFs can be even smaller than they are today, and a SFF is really all that anyone needs.

I've strayed a bit, so back to strictly storage for a moment. Smaller and faster is clearly a mantra that customers can appreciate - see sales of Raptors, and how WD has extended the Raptor's market segment in the face of surprising demand. This drive from Seagate could be a Raptor in a smaller box, which would be excellent.

What I find most interesting is that Seagate apparently has no interest in supporting a SATA interface on this drive, or in supporting the 44-pin (?) IDE interface used in laptops.

Share this post


Link to post
Share on other sites

Addressing a couple of points, it's no simple task to 'change' a SCSI drive into an IDE one. Drives are designed for SCSI, or for IDE, never for both. SCSI attracts more expensive components and more quality control, because companies are willing to pay more for a SCSI drive. SATA drives can be designed with cheaper components and more basic features, because most SATA customers are more concerned with price than with quality/features. It's not in a manufacturer's interests to build and qualify a drive for the SCSI market, then try to sell it in the IDE market. Apart from the (we?) enlightened few, most people won't see why it's so much money for so little space, when it supports no more features than it's 'value' SATA brethren.

Bearing all this in mind, what's the point of giving an SAS drive the ability to operate in 'crippled, SATA' mode? It would require extra electronics to support the different protocols, it wouldn't offer any extra value to the SCSI market, and it wouldn't offer much more than the competition in the SATA market. It would be more expensive to make than a straight SCSI drive, and as such would come into the "jack of all trades, master of none" category, as far as value went.

Controllers, OTOH, can only benefit from supporting a wider range of drives. They've got plenty of space for the extra electronics, and the electronics can form a higher proportion of total selling price.

And finally moving back on-topic, now that a 2.5" form factor exists, other manufacturers' SCSI and SATA teams may now start designing for it.

Share this post


Link to post
Share on other sites

Spod,

And finally moving back on-topic, now that a 2.5" form factor exists, other manufacturers' SCSI and SATA teams may now start designing for it.

This isn't the first 2.5" SAS Enterprise drive announcement. Fujitsu made a similar announcement not long ago. Of course it didn't get as much publicity. So there are already two competingmanufacturers with product lines in this class already.

Do well.

Jonathan Guilbault.

Share this post


Link to post
Share on other sites
full duplex data transfers

Isn't that only a concern if multiple devices share a single link in bandwidth intensive apps?

And why can SATA2 not do that, considering the physical interface is the same?

Please reread Spod's first post in this thread, your answer is there. Cost is king with SATA drives. Anything can be done but generally only the cost effective is done.

Free

Share this post


Link to post
Share on other sites
full duplex data transfers

Isn't that only a concern if multiple devices share a single link in bandwidth intensive apps?

And why can SATA2 not do that, considering the physical interface is the same?

Please reread Spod's first post in this thread, your answer is there. Cost is king with SATA drives. Anything can be done but generally only the cost effective is done.

Free

I've (of course) read his post. It's just silly that SATA HDDs can't be reliable in the eyes of lots of people.

Share this post


Link to post
Share on other sites
full duplex data transfers

Isn't that only a concern if multiple devices share a single link in bandwidth intensive apps?

And why can SATA2 not do that, considering the physical interface is the same?

Please reread Spod's first post in this thread, your answer is there. Cost is king with SATA drives. Anything can be done but generally only the cost effective is done.

Free

I've (of course) read his post. It's just silly that SATA HDDs can't be reliable in the eyes of lots of people.

They are quite reliable. I prefer the term less robust. :)

Free

Share this post


Link to post
Share on other sites

IBM has been desgning storage shelves for these for awhile, and this also allows their SCSI blades to be half the size.... pretty cool stuff. Quite the data density on the shelves.

Share this post


Link to post
Share on other sites
Addressing a couple of points, it's no simple task to 'change' a SCSI drive into an IDE one. Drives are designed for SCSI, or for IDE, never for both.

While this may have been true in the past it's not really true any more. SATA and SAS are electrically very similar(they were designed to be so) and thus would not require significant hardware changes to interoperate. It's more of a firmware issue than anything else...

Besides, you have to admit that the hardware differences between the parallel SCSI and SAS versions of the drive are quite significant but they are still doing it...

Apart from the (we?) enlightened few, most people won't see why it's so much money for so little space, when it supports no more features than it's 'value' SATA brethren.

It was "the enlightened few" that made the market for the Raptor(s). In the general market I will agree with you and freeborn that "Cost is king." However, this is not the case in the emerging market where the Raptor sits. In this market, people are willing to pay more for faster and more reliable drives, regardless of size...

Bearing all this in mind, what's the point of giving an SAS drive the ability to operate in 'crippled, SATA' mode?

Because I would buy it. The controller has always been the sticking point for SCSI with me. And yes, I know the argument that it is a one time cost, but this just isn't true in my case. As I purchase new drives, my older ones get shifted to my other computers or to family computers. So, until I start seeing SAS as a common interface in destop machines it is not an option.

It would require extra electronics to support the different protocols, it wouldn't offer any extra value to the SCSI market, and it wouldn't offer much more than the competition in the SATA market. It would be more expensive to make than a straight SCSI drive, and as such would come into the "jack of all trades, master of none" category, as far as value went.

Controllers, OTOH, can only benefit from supporting a wider range of drives. They've got plenty of space for the extra electronics, and the electronics can form a higher proportion of total selling price.

While the R&D might cost some money, the per-unit cost increase would be minimal. And it is completely irrelevant whether extra value to the SCSI market is added. What is important to the company is if it would sell more units and to this I can noanswer an unequivocal yes. See the Raptor for example...

As far as controllers go, there isn't much benefit to SATA support on SAS controller cards. Why would I waste valuable SAS ports to plug in an SATA drive when I can use the cheap SATA ports that come on practically any motherboard these days? The reason they did it is because it might be useful and it was cheap to accomplish because SAS and SATA are so similar.

The prime reason why this will never happen is because of all the techno-biggots in the IT sector that see SATA as some kind of taint. The HD companies fear that these people would buy a competitor's drive simply because it had the name SATA on it along with all the others.

-JoeTD

Share this post


Link to post
Share on other sites

I'm not familiar with the 74GB Raptor sales. Has it really taken off?

I don't think it has. Actually I think Raptor sales prove the opposite point: very few will pay a premium price. Would not Western digital rather attract *one* top PC manufacturer than the *entire* enthusiest market?

For example I recently bought a Hitachi 250GB 7k250 for ~$150US. I paid $105US for a 36GB Raptor when they first came out. Why would I pay $250 for a 74GB drive?

Is the Raptor worth 6x the cost of a 7k250? *shrug*

Market penetration:

I went shopping at the 1st Tier PC / Server sellers.

Only Dell (workstations) and Gateway (more limited) offered SCSI and Raptors in the Same space. However most systems that offered 7200rpm drives offered one or the other Raptor.

Let's say about a dozen systems offered Raptors and SCSI drives. Compare that to the hundreds of SCSI-based servers for sale. Clearly the money for such a drive is the true server market.

Would you pay $400 (let's assume it will be cheap) for a 2.5" 73GB drive?

I don't think that would fly in the desktop market.

I don't think it's elitism...it think it's realism. It is a difference between drives that are meant to run 24/7 under load and drives that are meant be idle most of the time - or even "off."

Not enough people will pay for the Savvio in the IDE-based Market. If Seagate decides to lose money and create a new market, fine. But they won't get their money back in a typical product lifecycle.

Besides my experiment with "Enterprise" IDE (36GB Raptor) is reporting bad blocks. Something I have know I have not experienced since the Barracuda debuted (the last one I remember was my "30MB" RLL drive before I moved to SCSI).

Regarding controller cards for SCSI:

It is the same as ATA. When you want to take advantage of a new hard drive you: 1) live with the slower speed of your motherboard's connector; or 2) buy a new controller card.

What's the difference? Seems to be the same with SCSI.

For example:

My U320 drives still works on my Adaptec 2940. Actually I can use my U320 drive on an Adaptec 2940U (Ultra SCSI) with little to no performance loss (80MB/sec) - what is that card now about 8-9 years old?. And they work fine on the U160 card I bought 3-4 years ago with no performance loss.

How is that different than people with ATA133 drives with ATA100 motherboards?

I will agree that cost when you *choose* to upgrade is bigger issue with SCSI. But do you really have to upgrade every time you buy a new drive?

Maybe WD just resurrected the Enterprise name for their lineup and didn't mean "Enterprise" performance,

Dogeared

8^)

Share this post


Link to post
Share on other sites

DogEared, you have some common sense :D

People spent double the amount on and extra 200MHz when buying Intel 3.0/3.2GHz CPU's

Some looneys will probably do the same for the extra cache on the EE versions...

People pay a large premium for slighly lower timing on the RAM, it get the ~3% speed increase for a cost far higher than 3% more....

It happens all over the place, the Raptop 74GB is 'the thing' and 'top technology' so some people will pay way over the odds for only a little bit more.

These companies know that and put a high price on early adopters while they can.

Share this post


Link to post
Share on other sites

Let me clear one thing up, I was a bit off-topic in my post as I was not discussing the Savvio in particular or the 2.5 inch form factor. I was just trying to correct misconceptions that there are any technical/cost reasons why a SCSI/SAS drive could not support SATA as well. The only significant barriers to this are marketing reasons.

DogEared:

I agree with you about the 74GB Raptor, I think WD has made a mistake here. But this proves my point, not disproves it. The opening market I mentioned is concerned with performance not storage space. The 74GB Raptor is too big and expensive for performance people and too small and expensive for storage people. However, I believe the original Raptor was an acceptable price for the performance gain and that a 36GB version of the current Raptor would be as well.

I currently operate under the 2 harddrive methodology. One small fast hard drive for my program/OS drive and one large cheaper drive for my data and rarely used programs. So my answer to your "is it worth it" question is yes and no.

Besides, regardless of how well the 74GB Raptor is accepted, the fact that it was released at all is proof enough that WD thought the original Raptor was successful.

Clearly the money for such a drive is the true server market.

Certainly, and I never argued otherwise. Still, it's not good business practice to not explore other markets for little to no extra cost...

It is a difference between drives that are meant to run 24/7 under load and drives that are meant be idle most of the time - or even "off."

That's one of the differences that I'm willing to pay for. The recent shifts to 1 year warranties has caused me much dismay and I would appreciate other options

Regarding controller cards for SCSI:

I believe you misunderstood my point, I was not talking about upgrading the controller card. Let me try to explain again. If I were ever to go the SCSI route, then my drives would have no alternate uses. When I upgraded drives, I could not give the old ones to family or friends(unless they had a SCSI controller). I could not move them to my other older machines(unless they had a SCSI controller).

So you see, unless I can count on family and friend's computers have SCSI or SAS capability, then those drives are useless to me. If, however, those same drives could be connected using SATA, then I can be guaranteed a "downgrade" path(for lack of a better word).

As an example, my Linux box, which I use for all my software development, is running off my 6 year old IBM 14GXP. If it had been SCSI, it would probably be in a spare parts box somewhere because I couldn't use it in any other system.

-JoeTD

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this