• 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.

Poll: Will SATA kill SCSI?

Chaotic42

Chaotic42

Lifer
I'm just curious about what those who are vastly more informed on the subject than myself think about SATA and SCSI. With the popularity of the 10K Raptor, and its ability to outperform the 15K Cheetah in many tasks, do you think that SATA will kill SCSI in the near future?

I don't think it will until atleast SATAII. SATA is just too slow and lacking in features right now, but its support is growing, and it seems like WD (who, IIRC has no SCSI interests) is leading the way.
 
SCSI will eventually go to Serial Attached SCSI.

It may even be possible to have the same drives on the same bus! <crosses fingers>

The fastest hosts are (and will always be) in the SCSI realm.

Kind of hard (currently) to have 30+ raptors hanging off a HBA.

Cheers!
 
The reasons SCSI will stay around a long time are RELIABILITY, device density per adapter and PERFORMANCE - the Cheetah hasn't been the king of the hill in SCSI for a good while now...
.bh.
 
SATA is close to top end SCSI in the 74 gig Raptor. But SCSI will be upgrading soon. Also SCSI drives still tend to have more reliablility and durable since they are used mostly in servers at running 24/7 365.
 
SATA and SCSI are marketed to different groups, home use vs. enterprise use. SCSI won't go way. Fibre Channel on the other hand doesn't seem to be gaining any ground on SCSI, (In non-SAN setups)
 
pffft SATA take out SCSI? not a chance!!! SCSI is still gaining ground and further advancing. SCSI is not going anywhere for along time esspecially at the enterprise level. I do think that SATA drives like the Raptor my make a dent in the enthusiest crowd who would normally go for SCSI but other than that no chance. Both are good at what they are ment for. My Raptor is awsome no doubt but it doesn't compete with many of the SCSI setups I have seen.
 
Can't remember the last time a product targetted for a different market killed another product. SATA will kill of SCSI when the pickup truck kills off the sports car.
 
SCSI is expensive.
It is expensive because the SCSI parts do a lot of work, and have a lot of features. SATA is getting some, and is becoming a popular NAS solution already, but SCSI will not die.

SATA II will give low-end and mid-range servers a cost-effective choice for drives and RAID arrays...but SCSI will be faster. It's not as much the protocol (though that helps) as it is that SCSI drives are made better, save for TWO, yes, only TWO, ATA drives. Except for the real expensive ATA ones, SCSI controllers are made better. You just can't beat that.
 
What I want to see is it SCSI loads signifacantly faster in gaming situations. If you check out Anand's article on Raptor vs 8Meg Cache drives, you'll notice the the Raptor does well in synthetic benchies...but there is not much of a difference (or not enough of a difference to pay an extra 70 dollars and give up potentially 50 gigs of space for 2 seconds of load time) in the actual games.

If SCSI preforms no better in gaming than the rest of the lot, then until SCSI gets "updated" Raptor does have the edge, because they also have 5 year warranties on them.

But until then, I'll hug my 200 and 80 gig SE Caviars and easily welcome an extra 250gigs of space for an extra 40ish dollars
 
Originally posted by: magomago
What I want to see is it SCSI loads signifacantly faster in gaming situations. If you check out Anand's article on Raptor vs 8Meg Cache drives, you'll notice the the Raptor does well in synthetic benchies...but there is not much of a difference (or not enough of a difference to pay an extra 70 dollars and give up potentially 50 gigs of space for 2 seconds of load time) in the actual games.

If SCSI preforms no better in gaming than the rest of the lot, then until SCSI gets "updated" Raptor does have the edge, because they also have 5 year warranties on them.

But until then, I'll hug my 200 and 80 gig SE Caviars and easily welcome an extra 250gigs of space for an extra 40ish dollars
Click to expand...
Um, no. Most SCSI drives also come with 5-year warranties, and perform similar to the Raptors on single-user tasks.
Also, you are not the target market for SCSI devices. People making database servers are. DB, web, X, etc. can bring any (current) ATA drive to its knees with only a dozen users or so. SATA II will be able to fix this, but most of the drives just aren't built well enough to survive a constant laod like SCSIs.
The gap will be bridged, but the markets are so different that there will be no need for ATA or SCSI to really compete. The trouble is that they are now, because dual and quad-CPU servers are becoming the mid-range, not the high-end, and businesses want to save some money.
Click to expand...
 
Originally posted by: Chaotic42
I'm just curious about what those who are vastly more informed on the subject than myself think about SATA and SCSI. With the popularity of the 10K Raptor, and its ability to outperform the 15K Cheetah in many tasks, do you think that SATA will kill SCSI in the near future?

I don't think it will until atleast SATAII. SATA is just too slow and lacking in features right now, but its support is growing, and it seems like WD (who, IIRC has no SCSI interests) is leading the way.
Click to expand...
Raptors beating 15k SCSI at anything ? Please provede details/benchmarks. I doubt it.
 
Yes, the newest Raptor has caught up to the slowest current gen 15k drive released 2 years ago in applications the SCSI drive wasn't optimized for. Not a tremendous accomplishment by any stretch of the imagination. The competitive balance will shortly swing completely back into SCSI's favor with the release of the already announced next generation of drives.
 
Originally posted by: Pariah
Yes, the newest Raptor has caught up to the slowest current gen 15k drive released 2 years ago in applications the SCSI drive wasn't optimized for. Not a tremendous accomplishment by any stretch of the imagination. The competitive balance will shortly swing completely back into SCSI's favor with the release of the already announced next generation of drives.
Click to expand...
Yes, I wouldn't call that a victory, but a good showing, yes.
 
You guys all have valid comments. SCSI isn't "going anywhere" per se, but SATA has hurt SCSI.

It's not really even SATA itself that has hurt SCSI. It's the performance ATA segment. Look at any SCSI drive, and I'll show you an ATA or SATA drive that is faster and cheaper. SCSI is now only practical for servers (and I say this as someone who has 9 SCSI drives in the two machines in front of him).

Power users used to get SCSI for their home systems all the time. I had plenty 10K drives back when the Cheetah was still a big deal, but I never had a SCSI RAID card because it was too expensive. Now you can get SATA RAID built into your motherboard cheap. ATA RAID has been on motherboards for a couple years. SCSI has become obsolete as far as most home users are concerned.

Yeah, the hardcore gamer and computer nut is a bit of a niche, but this translates to the workstation market, too. Places like HP and SGI will stop using SCSI in their workstations because people will eventually realize it's a waste of money. HP has already started using SATA workstations.

Look at the Raptor 74G vs. the 15K Atlas &amp; Cheetah. The Raptor is faster in several areas, and also significantly cooler and quieter. It's got a 5-year warranty. If you were building a new workstation, SCSI would be a waste of money.

When the CEO of a company comes home and sees that his son's $1000 PC blows the hell out of the $5000 workstations at the office, things will change.

SCSI is still the only answer in the server market, and it'll stay for a long time, but SATA is quickly taking the performance and workstation market. SATA won't kill SCSI, but it's already hurt it plenty.
 
Originally posted by: Pariah
Yes, the newest Raptor has caught up to the slowest current gen 15k drive released 2 years ago in applications the SCSI drive wasn't optimized for. Not a tremendous accomplishment by any stretch of the imagination. The competitive balance will shortly swing completely back into SCSI's favor with the release of the already announced next generation of drives.
Click to expand...

Well, I'm not saying it won, I'd still take the 15K. I'm just saying that I thought any 15K would trounce the Raptor.

Do you have any links to those new drives?
 
Originally posted by: Pariah
Yes, the newest Raptor has caught up to the slowest current gen 15k drive released 2 years ago in applications the SCSI drive wasn't optimized for. Not a tremendous accomplishment by any stretch of the imagination. The competitive balance will shortly swing completely back into SCSI's favor with the release of the already announced next generation of drives.
Click to expand...
Actually, it also was able to match the Fujitsu MAS 15k. That is of course, not including a single server bench, where the Raptor gets its balls handed to the Futjitsu on a silver platter.
However, yes, new generation drives. Always happens. Always will.
 
I love my SCSI setup and would never go back to IDE for a system/main drive.

At work, we have one small server in particular that is my "little pet." 5-drive RAID-5, 15K rpm SCSI drives...the thing just FLIES. It's our SAV/SMS box. It's always getting hit hard and never blinks twice. 😎
 
My first-generation 10k SCSI drive, a 9GB Quantum Atlas 10k, outperforms a Western Digital 800JB in a real-world I/O-intensive task, despite having about 1/3 the flat-out sprint speed (ok, possibly 1/2). StorageReview's benchmarks would almost certainly not reveal that, though. Kinda tells me something... StorageReview ain't God. 😉 They can say what they like, but reality > benchmarks. Now, if my job suddenly becomes the automated high-speed replay of StorageReview I/O traces, instead of actual real-world work, I'll demand that my employer buy me a 74GB Raptor, or more likely buy one myself :evil:

Everyone loves to pooh-pooh the SCSI drives as "server-optomized," great at I/O-intense stuff but poor at single-user stuff, but the reality is that when crunch time hits YOUR computer, that is when you'll notice the disk performance, or lack of it, the most. And that is where SCSI obliterates ATA completely. I fully realize that the other major obstacle for home users is that they want more capacity per dollar than SCSI has to offer.

Anyway, buy what suits your needs, guys. That's what I'll be doing. 😉 Cheetah 15k.4 and Fujitsu MAU have been announced. Cheetah 22k.1... hmmm 🙂 Pariah, any more info on the 22.5k drives you mentioned hearing about?
 
Originally posted by: Tostada
You guys all have valid comments. SCSI isn't "going anywhere" per se, but SATA has hurt SCSI.

It's not really even SATA itself that has hurt SCSI. It's the performance ATA segment. Look at any SCSI drive, and I'll show you an ATA or SATA drive that is faster and cheaper. SCSI is now only practical for servers (and I say this as someone who has 9 SCSI drives in the two machines in front of him).

Power users used to get SCSI for their home systems all the time. I had plenty 10K drives back when the Cheetah was still a big deal, but I never had a SCSI RAID card because it was too expensive. Now you can get SATA RAID built into your motherboard cheap. ATA RAID has been on motherboards for a couple years. SCSI has become obsolete as far as most home users are concerned.

Yeah, the hardcore gamer and computer nut is a bit of a niche, but this translates to the workstation market, too. Places like HP and SGI will stop using SCSI in their workstations because people will eventually realize it's a waste of money. HP has already started using SATA workstations.

Look at the Raptor 74G vs. the 15K Atlas &amp; Cheetah. The Raptor is faster in several areas, and also significantly cooler and quieter. It's got a 5-year warranty. If you were building a new workstation, SCSI would be a waste of money.

When the CEO of a company comes home and sees that his son's $1000 PC blows the hell out of the $5000 workstations at the office, things will change.

SCSI is still the only answer in the server market, and it'll stay for a long time, but SATA is quickly taking the performance and workstation market. SATA won't kill SCSI, but it's already hurt it plenty.
Click to expand...

I agree to some extent. There is no question that ATA/SATA is erroding SCSI's market in the lowend server and workstation markets. The reason for this, isn't so much that SCSI has taken a dive, but more of a case that for a lot of fields, ATA has simply become "fast enough" that the significant cost advantage it has will win it sales even if it isn't the king of performance. Another problem that has hit SCSI is the explosion of capacity requirements, especially for the home user which SCSI hasn't kept up with. SCSI used to be the king of performance and capacity. Now, ATA's below dirt cheap $/GB ratios for capacities that SCSI doesn't even offer keeps SCSI from ever gaining any market space in the consumer market again. Next generation SCSI will hit 147GB for 15k and 300GB for 10k which is great, but the cost will be too prohibitive to make them attractive to home users. And honestly, who needs 300GB's of industrial grade 10k hard drive capacity to store your pirated mp3, divx, and pr0n collections?

Everyone loves to pooh-pooh the SCSI drives as "server-optomized," great at I/O-intense stuff but poor at single-user stuff, but the reality is that when crunch time hits YOUR computer, that is when you'll notice the disk performance, or lack of it, the most.
Click to expand...

It's not that SCSI is poor at single user tasks, it's more that it is now only about comparable to the best ATA has to offer (Raptor) but at a greater cost. SCSI is still superior, and I wouldn't trade any of my SCSI drives for Raptors, but for most people the gains don't justify the additional cost. And again, I wouldn't use SCSI (or Raptors) for mass storage. The costs are simply too ridiculous to make it practical.

Kinda tells me something... StorageReview ain't God.
Click to expand...

No benchmarks from any site should be taken as gospel. They are only as useful as they apply to your own useage patterns. If you don't use the applications they test with, then the benchmarks mean nothing to you. If a video card kicks the crap out of another one in Far Cry, and you don't play Far Cry that benchmark is worthless to you as a buying criteria.

Cheetah 22k.1... hmmm Pariah, any more info on the 22.5k drives you mentioned hearing about?
Click to expand...

It was only a rumor. Nothing was ever substantiated. The recent mass announcements from Seagate mean if it is real, it is no where near release. If it ever does get released, it's practically a forgone conclusion that it will be released in the new 2.5" SAS interface which will take some time to gain acceptance.
 
SATA's reliability has to be proven in actual experience. It is not a question of speed. We know the good SCSI drives. We know which ones run for years, 24 hours a day, without problems and extremely low failure rates. There will not be a change unless SATA can match SCSI reliability. So far, I doubt SATA can prove itself. Too many. Too fast. Too soon. Too much greed. Too little on the engineering end.

and finally, why change?
 
Originally posted by: Zepper
The reasons SCSI will stay around a long time are RELIABILITY, device density per adapter and PERFORMANCE - the Cheetah hasn't been the king of the hill in SCSI for a good while now...
.bh.
Click to expand...

reliability and device density have little to do with bus choice, no? Also, performance largely depends on the HD, and in second turn on the controller?

It is possible that in 5 years the SATA logic will rival SCSI in terms of optimizing transfer (CQ, etc) and daisy-chaining, no? And the same disks can be made for SCSI / SATA, no? Would you still pay 3x as much only to be able to say "SCSI"?
 
Two totally different markets really. SCSI won't die anytime soon. Wasn't someone working on a serial scsi?

There are lots of companies out there that *gasp horror :Q etc* don't want the latest and greatest thing. Instead they rely on the tried and true. And that is? SCSI. 😉

I'd like to see SATA compete, but I doubt it will get into the server arena anytime soon.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top