SAS vs SATA RAID performance on multiple spindles
Hello,
I am trying to suggest options for a SAN storage array for a client. Customer has asked what % performance increase is there going from SATA to SAS drives in a RAID array, in order to justify the increased cost.
All the benchmarks I can find compare performance with multiple changed variables. I know there will be a performance increase, but I can't split out how much of the increase is directly attributed to just the SATA vs SAS technology change vs 10k RPM --> 15K RPM, RAID controller differences, etc.
So, presuming all things are equal (i.e. same RAID controller, same drive RPM, same RAID configuration), how much performance increase in average read and write throughput is expected by simply changing from SATA to SAS drives?
For clarification, a testbed example might be:
Adaptec RAID 31605 Controller
- Supports up to 16 (direct connect) SAS and/or SATA drives in any count
http://www.adaptec.com/en-US/products/Controllers/Hardware/sas/value/SAS-31605/
Barracuda ES.2 SATA 3.0-Gb/s 1-TB Hard Drive
http://preview.tinyurl.com/qzep6v
Barracuda ES.2 SAS 3.0-Gb/s 1-TB Hard Drive
http://preview.tinyurl.com/koyz2p
(yes, I know the SATA drive has double the cache of the SAS drive, closest comparision I could find)
If required, presume we are talking about a RAID 5 array w/ 8 drives.
I am trying to suggest options for a SAN storage array for a client. Customer has asked what % performance increase is there going from SATA to SAS drives in a RAID array, in order to justify the increased cost.
All the benchmarks I can find compare performance with multiple changed variables. I know there will be a performance increase, but I can't split out how much of the increase is directly attributed to just the SATA vs SAS technology change vs 10k RPM --> 15K RPM, RAID controller differences, etc.
So, presuming all things are equal (i.e. same RAID controller, same drive RPM, same RAID configuration), how much performance increase in average read and write throughput is expected by simply changing from SATA to SAS drives?
For clarification, a testbed example might be:
Adaptec RAID 31605 Controller
- Supports up to 16 (direct connect) SAS and/or SATA drives in any count
http://www.adaptec.com/en-US/products/Controllers/Hardware/sas/value/SAS-31605/
Barracuda ES.2 SATA 3.0-Gb/s 1-TB Hard Drive
http://preview.tinyurl.com/qzep6v
Barracuda ES.2 SAS 3.0-Gb/s 1-TB Hard Drive
http://preview.tinyurl.com/koyz2p
(yes, I know the SATA drive has double the cache of the SAS drive, closest comparision I could find)
If required, presume we are talking about a RAID 5 array w/ 8 drives.
leegclystvale
I don't know the answer to your question, but there is also a reliability issue with SAS v SATA.
ASKER
We plan on implementing a hot pluggable solution with online hot spares to address the reliability issue.
Cool. Interested in this thread and I hope there's a major performance gulf between the 2 or I've been well and truly shafted :o)
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Bottom line is: you get what you pay for.
So you're saying that SATA and SAS at 7200 packed into a RAID1, 5 or 10 array will perform equally at random workloads? Surely not! I was unde the impression that SATA is horrendously slow and cumbersome when used in this way.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ES.2 1TB SAS is just a SATA disk with a SAS interface, not really much difference. Reliability are the same and it doesn't have any of the benefits of a normal SAS 10k or 15K disk.
BTW, Gary, aerial density is the number of aerials in a given area, I think you mean areal density ;)
BTW, Gary, aerial density is the number of aerials in a given area, I think you mean areal density ;)
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
"... I think you mean areal density ;) " ==> :-) As I'm sure you know, that was a typo, not a misunderstanding. (since corrected)
> (since corrected)
Editing your posts is not really in order unless we can *all* edit our posts, you spoil the joke by editing in that areal isn't in our dictionaries and it isn't a typo - it's a new word.
In some ways 10k SATAs like Cheetahs and 7.2k SAS like this ES.2 also need new words, like HP use FATA to describe their low performance disks attached by a mismatched performance-wise f/c interface.
Editing your posts is not really in order unless we can *all* edit our posts, you spoil the joke by editing in that areal isn't in our dictionaries and it isn't a typo - it's a new word.
In some ways 10k SATAs like Cheetahs and 7.2k SAS like this ES.2 also need new words, like HP use FATA to describe their low performance disks attached by a mismatched performance-wise f/c interface.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for the responses so far.
To elaborate and clarify:
1) I understand that a pure SATA to SAS comparison is somewhat artificial. The point of the question is to try and avoid clouding the issue with rotational speeds, 2.5" vs 3.5" form factors, etc. I want to avoid false conclusions, i.e. if a red racecar and a blue bus compete in a race and the red racecar won, it would be specious at best to conclude that red vehicles are faster.
2) for purposes of this test case, presume that we are trying to run transactional data (Exchange email stores, SQL databases) off the array. "Drag and drop" performance is not my main focus for this question.
3) Our clients are very small businesses (typically < 25 employees). Common sense says that a 1000 man company should not use SATA drives for an email store. Likewise, most would agree that a 2 man company can live with SATA drives in a RAID 1. How about 2 heavy users? 5 light users? 10? 50? The line where SATA drives are no longer acceptable is a quality assessment, based on the user experience (opinion-based) and performance statistics (fact-based). I am looking for quantitative data to help my clients make qualitative assessments.
4) The purpose of asking for a "% increase" is twofold:
a) I can easily calculate the % increase in price going from SATA drives to SAS drives. I can also find price increase percentages for other hard drive performance factors (i.e. rpm, cache size, etc).
b) I can compare the ROI of moving from SATA to SAS drives, compared with other hard drive improvements. Furthermore, I can compare the ROI of storage improvements vs other focus areas (e.g. improved network switches, CPU power, RAM, etc)
5) The affordability of a solution is ultimately a business decision, not a technical one. We view our responsibility to provide accurate data on the costs and results of their choices. Specifically, we have 5 man engineering clients who would call $6k for SAS drives 'petty cash' and worth doing. We also have clients with 80 staff who couldn't justify an extra 1.5 TB desktop drive in a whitebox server for $150. We prefer to say $$ for solution A, $$$ for solution B with an estimated % increase for the additional $, and let the client choose.
I will continue to leave this question open for now, as it seems like there is intelligent discussion and interest on the topic.
To elaborate and clarify:
1) I understand that a pure SATA to SAS comparison is somewhat artificial. The point of the question is to try and avoid clouding the issue with rotational speeds, 2.5" vs 3.5" form factors, etc. I want to avoid false conclusions, i.e. if a red racecar and a blue bus compete in a race and the red racecar won, it would be specious at best to conclude that red vehicles are faster.
2) for purposes of this test case, presume that we are trying to run transactional data (Exchange email stores, SQL databases) off the array. "Drag and drop" performance is not my main focus for this question.
3) Our clients are very small businesses (typically < 25 employees). Common sense says that a 1000 man company should not use SATA drives for an email store. Likewise, most would agree that a 2 man company can live with SATA drives in a RAID 1. How about 2 heavy users? 5 light users? 10? 50? The line where SATA drives are no longer acceptable is a quality assessment, based on the user experience (opinion-based) and performance statistics (fact-based). I am looking for quantitative data to help my clients make qualitative assessments.
4) The purpose of asking for a "% increase" is twofold:
a) I can easily calculate the % increase in price going from SATA drives to SAS drives. I can also find price increase percentages for other hard drive performance factors (i.e. rpm, cache size, etc).
b) I can compare the ROI of moving from SATA to SAS drives, compared with other hard drive improvements. Furthermore, I can compare the ROI of storage improvements vs other focus areas (e.g. improved network switches, CPU power, RAM, etc)
5) The affordability of a solution is ultimately a business decision, not a technical one. We view our responsibility to provide accurate data on the costs and results of their choices. Specifically, we have 5 man engineering clients who would call $6k for SAS drives 'petty cash' and worth doing. We also have clients with 80 staff who couldn't justify an extra 1.5 TB desktop drive in a whitebox server for $150. We prefer to say $$ for solution A, $$$ for solution B with an estimated % increase for the additional $, and let the client choose.
I will continue to leave this question open for now, as it seems like there is intelligent discussion and interest on the topic.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SCSI = SAS = FC/Fibre Channel
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Ooops. My bad - I was looking at the wrong line...
I just thought I would put my 2c worth in.
I thought would comment on a few things.
Firstly thought it was interesting that wasnt until a number of posts that anyone commented on the RAID array type and the overhead that different options create.
Interestingly you are looking at the 3 series Adaptec cards. I am not sure why you are not looking at the 5 series because of the increased performance that is available with them.
From all the discussion I think there is one thing that everyone can agree with "1 size doesnt fit all!"
It depends on so many variations.
I think to get some specific advise you need to ask a specific question.
One size does not fit all.
I thought would comment on a few things.
Firstly thought it was interesting that wasnt until a number of posts that anyone commented on the RAID array type and the overhead that different options create.
Interestingly you are looking at the 3 series Adaptec cards. I am not sure why you are not looking at the 5 series because of the increased performance that is available with them.
From all the discussion I think there is one thing that everyone can agree with "1 size doesnt fit all!"
It depends on so many variations.
I think to get some specific advise you need to ask a specific question.
One size does not fit all.
He actually says in the question that he's talking about a SAN so I think the Adaptec card is irrelevant.
ASKER
I chose the RAID adapter simply as an example of a mixed SATA/SAS card.
Based upon meyersd and garycase's comments above, am I correct in concluding that a SATA RAID with enough drives can outperform a SAS RAID with less drives?
i.e. presuming SATA 7.2k @ 60 IOPS and SAS 10K @ 140 IOPS, would a 6 drive SATA RAID 1 outperform a 2 drive SAS RAID 1 by about 10 ~ 25%?
Based upon meyersd and garycase's comments above, am I correct in concluding that a SATA RAID with enough drives can outperform a SAS RAID with less drives?
i.e. presuming SATA 7.2k @ 60 IOPS and SAS 10K @ 140 IOPS, would a 6 drive SATA RAID 1 outperform a 2 drive SAS RAID 1 by about 10 ~ 25%?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Sorry, I should have been more clear: my focus is transactional data in all my comments, including the theoretical 6 drive vs 2 drive scenario. Restated:
Presuming a 7.2k HDD @ 60 IOPS and a 10k HDD @ 140 IOPS, would a 6 drive RAID 1 w/ 7.2k drives outperform a 2 drive RAID 1 w/ 10k drives, for transactional data e.g. Exchange email store?
I realize that I am drifting away from my initial question. Should this question re: 6 vs 2 drives be in a new post? I will assign points to this topic later today.
Presuming a 7.2k HDD @ 60 IOPS and a 10k HDD @ 140 IOPS, would a 6 drive RAID 1 w/ 7.2k drives outperform a 2 drive RAID 1 w/ 10k drives, for transactional data e.g. Exchange email store?
I realize that I am drifting away from my initial question. Should this question re: 6 vs 2 drives be in a new post? I will assign points to this topic later today.
I think that constitutes a new post - although you're still referring back to HDD performance this is more of a RAID configuration discussion...that and this post is unending.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
It seems to be a sales pitch for a PCIe Adaptec RAID controller to plug into a server? Not something that is very useful in a SAN since it does not provide full redundancy.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
>Based upon meyersd and garycase's comments above, am I correct in concluding that a SATA RAID with enough drives can outperform a SAS RAID with less drives?
Of course it can, a million horses can plough a field faster than one diesel driven tractor.
Of course it can, a million horses can plough a field faster than one diesel driven tractor.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Your first question is a doozy, thebuyguy... :-)
Excuse my ignorance as I know little about SAN's other than they are expensive, but what type of SAN hardware are you looking at installing these drives into and how will these be connected to the servers - I assume thru fibre-channel. Also the little I have seen with SAN storage units is that they are pretty expensive and I would have thought for most <25user business would have been price prohibative.
Thanks for all those much more experienced in this field that have been commenting as it is very interesting some of the information.
Thanks for all those much more experienced in this field that have been commenting as it is very interesting some of the information.
Gary - just about some of the enterprise features in the WD HDD, is this just with the latest RE3 generation of products or has this been around before that?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Thanks Gary. I have both 1TB RE3 WD SATA & 1TB SAS ES HDD so I will see if I can test this and let you know how I go.
ASKER
Thank you very much -- very thorough discussion.