Now test a RAID card that isn't software RAID or PCI-E bus limited, please. ;) Admittedly I've not tried RAID off on ICH10R, but my experiences with software RAID on ICH9 are not good in the slightest.
The TX4650 is an outdated, entry-level SAS RAID controller for those looking to add basic SAS (_not_ SATA) connectivity to box, with bare-minimum RAID capability and performance.
The overhead brought with SAS support has been worsens SATA performance, and it is well known that Intel's storage contollers excel with SSDs. The Tx4650 does not perform parity calculations 'on its own CPU' and leaves most of the RAID management to software. I just don't understand why you would choose such a controller for a build like this, when a much better-performing, but equivalently-priced, SATA controller would be much more useful suitable candidate.
This could have been a good review, had you looked at a suitable controller. The conclusion would probably have been much the same - that for a stand-alone NAS box, there is no advantage to a dedicated controller because of the greater power consumption, and because software RAID does not provide such a performance deficit unless it is measured in terms of free CPU time - a moot point for a NAS box, whose sole purpose _is_ storage.
In short, the choice of RAID controller is a poor one considering the application. A good storage review would test the product in question with consideration of its intended target platforms. Bit-tech's journalism has been going downhill for some time, but this really is a new low.
Originally Posted by karbonKid Wow. What a useless review.
The TX4650 is an outdated, entry-level SAS RAID controller for those looking to add basic SAS (_not_ SATA) connectivity to box, with bare-minimum RAID capability and performance.
The overhead brought with SAS support has been worsens SATA performance, and it is well known that Intel's storage contollers excel with SSDs. The Tx4650 does not perform parity calculations 'on its own CPU' and leaves most of the RAID management to software. I just don't understand why you would choose such a controller for a build like this, when a much better-performing, but equivalently-priced, SATA controller would be much more useful suitable candidate.
This could have been a good review, had you looked at a suitable controller. The conclusion would probably have been much the same - that for a stand-alone NAS box, there is no advantage to a dedicated controller because of the greater power consumption, and because software RAID does not provide such a performance deficit unless it is measured in terms of free CPU time - a moot point for a NAS box, whose sole purpose _is_ storage.
In short, the choice of RAID controller is a poor one considering the application. A good storage review would test the product in question with consideration of its intended target platforms. Bit-tech's journalism has been going downhill for some time, but this really is a new low.
I could have done with a decent review of the SRCSATAWB before i went and spent £250 on it.....
turns out although it does have good hardware raid 5/6 performance, it just isnt very flexible. You cant use jbod, and ive yet to see any option of expanding an array.
I wouldn't say "Bit-tech's journalism has been going downhill for some time" - I think that's unfair and unjustified - but in fairness, it does seem a rather bizarre idea to do a full and comprehensive review of a product that is at least three years old.
I say 'at least', I don't know how old it actually is, but the fact-sheet for it on the Promise website is dated 29th August 2007 - and that could be an updated fact-sheet for all I know. It mentions it's compatible with Windows XP, but there's no mention even of Vista - released Jan 2007!
Sorry guys. Must have got up on the wrong side of the bed this morning (perhaps because my cat decided to use my right eyelid to test the sharpness of his claws, some time last night).
I don't understand why you reviewed an old software RAID card... Cards like this have FAIL writ large all over them. They will almost certainly be worse than a i7 powered ICH10 set up.
Why not review up to date hardware RAID cards?
This review will just make the uninitiated think that RAID cards are crap, especially as the one you reviewed is massively overpriced and useless.
well I sure would not waste money on a RAID card in a NAS. Just go with software RAID, which offers boatloads of more flexibility than any hardware RAID can give.
The sole purpose of a NAS BOX as KarbonKid said is to do NAS tasks.
Originally Posted by karbonKid Wow. What a useless review.
The TX4650 is an outdated, entry-level SAS RAID controller for those looking to add basic SAS (_not_ SATA) connectivity to box, with bare-minimum RAID capability and performance.
The overhead brought with SAS support has been worsens SATA performance, and it is well known that Intel's storage contollers excel with SSDs. The Tx4650 does not perform parity calculations 'on its own CPU' and leaves most of the RAID management to software. I just don't understand why you would choose such a controller for a build like this, when a much better-performing, but equivalently-priced, SATA controller would be much more useful suitable candidate.
This could have been a good review, had you looked at a suitable controller. The conclusion would probably have been much the same - that for a stand-alone NAS box, there is no advantage to a dedicated controller because of the greater power consumption, and because software RAID does not provide such a performance deficit unless it is measured in terms of free CPU time - a moot point for a NAS box, whose sole purpose _is_ storage.
In short, the choice of RAID controller is a poor one considering the application. A good storage review would test the product in question with consideration of its intended target platforms. Bit-tech's journalism has been going downhill for some time, but this really is a new low.
I'm sorry, did you actually read the review?
We said "uses THE CPU" as in - the main system CPU, not its own.
Promise advertise it as SAS AND SATA - most home users will use it with SATA. We tested it with SATA. I don't see why a product can't be built with a firmware to benefit both - I would at least expect as much, certainly not crippling SATA at expense of niche SAS.
This review was not dedicated to a NAS box build - it was merely discussed as a possible use for the product, if people wanted to expand their boxes at not much expense. As explained in the opening paragraphs we were given it for another proof of concept article, by Promise, and then thought - "hey guys, just FYI, this is how it performs". We don't do many RAID cards - we just figured you maybe interested in the numbers and something a little different.
We were interested to see how it compared paying £100 for a card versus out the box, onboard RAID. "Buying a RAID card" is usually pimped as being much better, much safer than your onboard m'larky, but clearly from the results that's not the case. Our conclusion is simply - don't bother, and don't assume it's better just because "it's not onboard RAID".
Pook - to the uneducated THESE kind of cards are exactly what they buy, and as we found, yes, it's ****. I hope they conclude that they should spend more or just use onboard or software RAID - NOT avoid all of them altogether. Even if its i7 or whatever - the ICH10 core has been the same since P45: it's in P55 and also partners X58 too - hell it's even more or less the same core all the way back to ICH8, only the software has improved massively!
It would be nice for you to review other add in £100 RAID cards and see how they perform, perhaps there is a difference in quality / manufacturer. Perhaps try an Adaptec RAID 1430SA.
The fact still remains that SAS support does introduce a performance overhead, that's just the way it is due to the extended feature set of SAS. The fact is that with a card like this, you're not paying £100 for RAID (Not least because _it_ doesn't even do RAID), you're paying £100 for SAS.
A look at one of LSI's newer low-end SATA RAID controllers (based on the product line they inherited from AMCC/3Ware), or of low-end Highpoint RocketRAIDs, etc. would make for a much more relevant review and much more convincingly dispell the notions you speak of.
I'll leave it at that, and I apologize once again for my tone earlier.
We don't do many RAID cards - we just figured you maybe interested in the numbers and something a little different.
Would be interesting to see reviews of some "proper" raid cards as Raid is getting more important these days due to the small capacity of ssds essentially promoting the use of Raid 0 and with mechanical hdds being dirt cheap redundancy like for example in Raid 5 being a attractive option as well.
just my 2 cents.
I would like to second the comments about this review being rather bizarre. To be honest I only read through it to try and figure out why you testing a soft-raid, overpriced (for SATA) card. I would love to see more Host Controller card reviews. I would like to see reviews of more expensive hardware-raid SATA or SATA/SAS multiport cards : looking at both attached SAS and SATA drives for performance (I/O performance). Plus support in "true" server operating systems e.g.: Solaris, OpenBSD (FreeNAS) and Linux variants.
I know, I know it'll never happen. But this would be very useful before handing over big wads of hard-earned cash only to get crap driver support/ performance...
but the card is crap if not all 1x based cards are due to limited bandwidth and software RAID or its limited hardware RAID as some are hardware based just the CPU on them are just to slow to handle tasks (it was good at 1-2 RAID 5 tests that seem very odd)
for £100 it is an waste of money for an 4 port RAID card that does little should of gotten an 1, how good was its Disk to disk speeds over the card it self be interesting if it was able to copy at 100MB/s form disk to disk on the same card (total speed combined should be 200-220MB/s if not it sucks very much)
Pook - to the uneducated THESE kind of cards are exactly what they buy, and as we found, yes, it's ****. I hope they conclude that they should spend more or just use onboard or software RAID - NOT avoid all of them altogether.
Comments 1 to 20 of 20
ReplyThe TX4650 is an outdated, entry-level SAS RAID controller for those looking to add basic SAS (_not_ SATA) connectivity to box, with bare-minimum RAID capability and performance.
The overhead brought with SAS support has been worsens SATA performance, and it is well known that Intel's storage contollers excel with SSDs. The Tx4650 does not perform parity calculations 'on its own CPU' and leaves most of the RAID management to software. I just don't understand why you would choose such a controller for a build like this, when a much better-performing, but equivalently-priced, SATA controller would be much more useful suitable candidate.
This could have been a good review, had you looked at a suitable controller. The conclusion would probably have been much the same - that for a stand-alone NAS box, there is no advantage to a dedicated controller because of the greater power consumption, and because software RAID does not provide such a performance deficit unless it is measured in terms of free CPU time - a moot point for a NAS box, whose sole purpose _is_ storage.
In short, the choice of RAID controller is a poor one considering the application. A good storage review would test the product in question with consideration of its intended target platforms. Bit-tech's journalism has been going downhill for some time, but this really is a new low.
Would be interesting to see some reviews of better RAID cards as well.
ooooooooooooOOOOOOOOOOOOOOooooooooooohhhhhhhh!
turns out although it does have good hardware raid 5/6 performance, it just isnt very flexible. You cant use jbod, and ive yet to see any option of expanding an array.
I say 'at least', I don't know how old it actually is, but the fact-sheet for it on the Promise website is dated 29th August 2007 - and that could be an updated fact-sheet for all I know. It mentions it's compatible with Windows XP, but there's no mention even of Vista - released Jan 2007!
http://www.promise.com/media_bank/Download%20Bank/Datasheet/FastTrak%20Family.pdf
Would you guys do a full review of a graphics card, motherboard, or anything else that's over three years old and conclude it was rubbish?
I mean I understand the curiosity you mentioned while you had the thing to hand, but like I said it just seems, well, bizarre...
Why not review up to date hardware RAID cards?
This review will just make the uninitiated think that RAID cards are crap, especially as the one you reviewed is massively overpriced and useless.
i thought the exact same thing
The sole purpose of a NAS BOX as KarbonKid said is to do NAS tasks.
I'm sorry, did you actually read the review?
We said "uses THE CPU" as in - the main system CPU, not its own.
Promise advertise it as SAS AND SATA - most home users will use it with SATA. We tested it with SATA. I don't see why a product can't be built with a firmware to benefit both - I would at least expect as much, certainly not crippling SATA at expense of niche SAS.
This review was not dedicated to a NAS box build - it was merely discussed as a possible use for the product, if people wanted to expand their boxes at not much expense. As explained in the opening paragraphs we were given it for another proof of concept article, by Promise, and then thought - "hey guys, just FYI, this is how it performs". We don't do many RAID cards - we just figured you maybe interested in the numbers and something a little different.
We were interested to see how it compared paying £100 for a card versus out the box, onboard RAID. "Buying a RAID card" is usually pimped as being much better, much safer than your onboard m'larky, but clearly from the results that's not the case. Our conclusion is simply - don't bother, and don't assume it's better just because "it's not onboard RAID".
Pook - to the uneducated THESE kind of cards are exactly what they buy, and as we found, yes, it's ****. I hope they conclude that they should spend more or just use onboard or software RAID - NOT avoid all of them altogether. Even if its i7 or whatever - the ICH10 core has been the same since P45: it's in P55 and also partners X58 too - hell it's even more or less the same core all the way back to ICH8, only the software has improved massively!
The fact still remains that SAS support does introduce a performance overhead, that's just the way it is due to the extended feature set of SAS. The fact is that with a card like this, you're not paying £100 for RAID (Not least because _it_ doesn't even do RAID), you're paying £100 for SAS.
A look at one of LSI's newer low-end SATA RAID controllers (based on the product line they inherited from AMCC/3Ware), or of low-end Highpoint RocketRAIDs, etc. would make for a much more relevant review and much more convincingly dispell the notions you speak of.
I'll leave it at that, and I apologize once again for my tone earlier.
I would do more if I could get some contacts. I'll work on it :)
Would be interesting to see reviews of some "proper" raid cards as Raid is getting more important these days due to the small capacity of ssds essentially promoting the use of Raid 0 and with mechanical hdds being dirt cheap redundancy like for example in Raid 5 being a attractive option as well.
just my 2 cents.
I know, I know it'll never happen. But this would be very useful before handing over big wads of hard-earned cash only to get crap driver support/ performance...
for £100 it is an waste of money for an 4 port RAID card that does little should of gotten an 1, how good was its Disk to disk speeds over the card it self be interesting if it was able to copy at 100MB/s form disk to disk on the same card (total speed combined should be 200-220MB/s if not it sucks very much)
:) forgot to press post
Fair point. Apologies.
-
« Previous
-
1
-
Next »
Discuss in the forums