I have to completely disagree with this. There is nothing slower in a computer/server than the hard drives. Can other portions be bottlenecks, absolutely, but you aren't typically running out of CPU or even memory on a 32-bit system. why do you think most companies go with 2cpu systems vs 4? Becaues CPU isn't the bottleneck (typically) in a 32-bit system. As well, your arguement falls flat because now we are talking about duo-core dies so those blades, 1U's, etc are now 4way boxes and can be taken advantage of in 64-bit implementations.
I'm sorry there is no good arguement for going SATA over SCSI in a TS environment. It's short sighted and your shooting yourself in the foot before you even get off the ground.
Jeff Pitsch Microsoft MVP - Terminal Server Provision Networks VIP
Forums not enough? Get support from the experts at your business http://jeffpitschconsulting.com
But will it be the smallest bottleneck, might the # of CPUs (specially in Blade systems) not be another potential bottleneck ?
I was going chime in last week regarding the warranty, maybe Compaq or some other only give you a 1 year warranty, but most drives these days (even PATA) have 3 or 5 year warranty.
And it is true what Amer said, we had a notorious SCSI HD failure rate on our IBM X350, X360 and X365 servers--blame Hitachi.
------------------------------ *From:* thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] *On Behalf Of *Jeff Pitsch *Sent:* Sunday, October 01, 2006 8:16 PM *To:* thin@xxxxxxxxxxxxx *Subject:* [THIN] Re: SATA drives
Performance drop. Price is fine, reliablity maybe, but performance is much worse than SCSI. you are putting in a bottleneck that is unneeded. You aren't supposed to create your own bottleneck in a TS environment.
Jeff Pitsch Microsoft MVP - Terminal Server Provision Networks VIP
Forums not enough? Get support from the experts at your business http://jeffpitschconsulting.com
On 9/30/06, Amer Karim <amerk@xxxxxxxxxxxxxxxx> wrote: > > I agree with you wrt to SATA being appropriate dependant on what they > are going to be used for, and how. However, regarding published failure > rates and reliability figures - as far as I'm concerned, they're > meaningless. We went through a wonderful period of about 2 years where we > were experiencing a failure rate of about 80% (note the missing preceding > decimal) on brand new U320 SCSI drives (from various manufacturers) - almost > every one of them brand new, and most within 6 to 12 months of use; > including one reputation breaking case where every single drive for a new > SBS server, 6 drives in RAID-1 and RAID-5 w/ hot-spare, failed the burn in; > we RMA'd those, and when we got the replacements, all of those failed > again. By then we had swapped out the system board, power supplies, gone > through 4 different RAID controllers because the Seagate chaps were > convinced the drives were being blown by the surrounding hardware. We > decided to switch to IBM drives instead - and had half of them die on > us. Turned out the problems were being caused by a bug in the drives' > firmware. We're still seeing a failure rate on U320 SCSI drives, both 10K > and 15K flavours, which is far greater than it used to be 3 years ago - > about 1 in 20 on average, and we've RMA'd more SCSI drives in the last 3 > years than we did in the preceding 10. > > Thus, IMO, figures on failure rates and reliability are moot - one bug > in a firmware revision and that much vaunted integrity and reputation is mud > as far as a client is concerned. I agree there are some applications where > SCSI performance is still a necessity - but I no longer consider them the > holy grail, and if a SATA drive goes south I can replace it in however long > it takes me to get to the client's site - they're cheap enough, and > available enough. Keeping my clients systems up and running is what they > pay me for - and redundancy does a far better job of that than hardware > 'reliability' and 'failure' figures. > > I will also state that up until 2 years ago, I would have, and did, walk > away from any client who did not want to spend the money on putting SCSI > drives in their servers. I'm confident enough in the newer SATA/SAS > technologies that I now consider them viable, and in certain cases > preferable, alternatives to SCSI. > > Regards, > Amer Karim > Nautilis Information Systems > > > -----Original Message----- > From: thin-bounce@xxxxxxxxxxxxx [mailto: thin-bounce@xxxxxxxxxxxxx] On > Behalf Of Roger Riggins > Sent: September 29, 2006 10:30 PM > To: thin@xxxxxxxxxxxxx > Subject: [THIN] Re: SATA drives > > SATA can be an alternative for SCSI if you want cheaper, but can accept > slower and less reliable storage. We found a home for them in our D2D2T > solution, but those specs were acceptable for that project. Everyone's > requirements are different for every project. Personally, if I'm > responsible for the equipment or my reputation is on the line, then I'm > going to recommend what I believe to be the best. If I have to > compromise integrity for price, then I make sure that management > understands that. I keep an "I told you so" in my back pocket. :) > > Check this out-- > > Seek times: > > SAS Barracuda ES: 8.5/9.5 > SAS NL35: 8.0/9.0 > > SCSI/SAS Cheetah 15k: 3.5/4.0 > SCSI/SAS Savvio 10k.2: 3.8/4.4 > > > Sustained transfer rate: > > SAS Barracuda ES: up to 78 Mbytes/sec > SAS NL35: up to 65 Mbytes/sec > > SCSI/SAS Cheetah 15k: up to 125 Mbytes/sec > SCSI/SAS Savvio 10k.2: up to 85 Mbytes/sec > > > Annualized Fail Rate at 24x7 operation: > > SAS Barracuda ES: .73% > Others: not listed, probably worse since the Barracuda is supposed to be > > their most reliable SATA or probably not rated for 24x7 operation > > SCSI/SAS Cheetah 15k: .62% > SCSI/SAS Savvio 10k.2: .55% > > > So from the numbers, it's safe to say that the SCSI/SAS seek times are > almost half of SATA. Additionally, the above SATA drives are up to 25% > more likely to fail than a SCSI/SAS drive. That's if you get these new > ones that are supposed to be more reliable than the others! > > Here's an interesting link from the makers of the Barracuda. ;) > > http://www.seagate.com/products/interface/sata/targetapp.html > > So the bottom line is that SATA is a viable alternative for SCSI/SAS, > but mostly for specific solutions/projects or very small shops. > > Good luck, > > Roger Riggins > Network Administrator > Lutheran Services in Iowa > w: 319.859.3543 > c: 319.290.5687 > http://www.lsiowa.org > > > > -----Original Message----- > From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On > Behalf Of Amer Karim > Sent: Friday, September 29, 2006 6:39 PM > To: thin@xxxxxxxxxxxxx > Subject: [THIN] Re: SATA drives > > The Seagate 3GB/s SATA drives (Barracuda ES) lines have a 5-year > warranty - and, for the price, I can put 8 of those in a server with > RAID-10 and RAID-5 with 2 hot-spares for a fraction of the cost of SCSI > for equivalent capacity. In other words, I'd have to disagree with the > comments about SATA not being a viable alternative to SCSI/SAS. And > throw in an SAS RAID controller, and you've made the migration to SAS > drives down the road a fairly simple thing as well. The SATA disks > being referred to in those articles are older tech and better suited for > desktop computers, rather than servers - IMHO. > > Regards, > Amer Karim > Nautilis Information Systems > > > > <b>Lutheran Services in Iowa Confidentiality Notice > ==================================================================</b> > <red>The information contained in this communication may be > confidential, > is intended only for the use of the recipient(s) named above, and > may be legally privileged. If the reader of this message is not the > intended recipient, you are hereby notified that any dissemination, > distribution, or copying of this communication, or any of its > contents, is strictly prohibited. If you have received this > communication in error, please return it to the sender immediately > and delete the original message and any copy of it from your computer > system. If you have any questions concerning this message, please > contact the sender.</red>½IRn‰¹(r)¢´zjžz¶z–†ÿÁz¶ƒ² >