Product Qualification: Intel S1200BTS Motherboard

Written on November 30, 2011 by Matt Bach
Share:

Introduction

Intel S1200BTSThe Intel S1200BTS (based on the C202 chipset) is a micro ATX server motherboard designed for small to medium-sized business applications. Puget Systems is already offering another Micro ATX server motherboard - the Asus P8B-M based on the C204 chipset - but Intel motherboards often make a better choice for server platforms.. This is not due to any sort of quality issues with other brands, but simply that Intel does not add any additional features that are not already present on the chipset and is very proactive about becoming certified for server-class software. While this may mean that the usability of the motherboard is not as intuitive (no UEFI BIOS, firmware updates must be done from a command line) stability is the number one priority for a server and the fewer features on a motherboard, the fewer things that can fail.

Since we are considering replacing the Asus P8B-M with this motherboard, lets first go through the specifications for each motherboard before we take a closer look at the Intel S1200BTS:

  Intel S1200BTS Asus P8B-M  
Socket 1155 1155
Form Factor Micro ATX Micro ATX
Chipset C202 C204
Networking Intel 82574L, Intel 82579 Intel 82574L
Audio Device N/A N/A
Remote Management N/A ASMB5-iKVM
Memory Support  
Type DDR3-1333 DDR3-1333
# of Slots 4 4
Maximum Allowed 32GB 32GB
Onboard Video  
Processor Type Intel SM712 Aspeed AST2050
Onboard RAM Shared 16MB
Ports VGA VGA
External Connections  
USB 2.0 4 2
PS/2 N/A Keyboard & Mouse
LAN (RJ45) 2 2
Serial (COM) 1 1
Internal Connections  
USB 2.0 1 2
SATA 3Gb/s 6 (Intel C202) 4 (Intel C204)
SATA 6Gb/s N/A 2 (Intel C204)
RAID Support 0,1,5,10 (0,1,10 only with LSI Controller) 0,1,5,10 (0,1,10 only with LSI Controller)
Expansion Slots  
Slot 1 PCI Express 2.0 x8 in a x16 size PCI Express 2.0 x8
Slot 2 PCI Express 2.0 x8 PCI Express 2.0 x16
Slot 3 PCI Express 2.0 x8 PCI Express 2.0 x8
Slot 4 PCI PCI


While the chipsets themselves are different, this is not actually as big of a deal as it sounds. In fact, the only major difference between the C202 and C204 chipsets is that the C204 supports 4 SATA 3Gb/s and 2 SATA 6Gb/s ports while the C202 chipset only supports 6 SATA 3Gb/s ports. While this does limit the speed of the hard drives used with the Intel S1200BTS, it really only comes into play when using super-fast SSD drives.

Both of these motherboards also feature a built-in LSI RAID controller. Usually, onboard RAID just uses the same controller that is used for single-disk configurations. We have found that this RAID functionality is often added as more of an afterthought, and many times has a plethora of stability issues. By integrating an LSI RAID controller into the motherboard, Intel has finally provided onboard RAID functionality that should be just as stable and reliable as a dedicated RAID card. This is definitely something we will be testing further on in this article.

The biggest advantage the Asus P8B-M has over the Intel S1200BTS is the remote management capability. With the P8B-M, you can purchase and install the Asus ASMB5-iKVM module to get a plethora of remote management options include remote power on/off and remote desktop at the BIOS level. Intel does have a Remote Management Module available, but is currently only supported on the Intel S1200BTL.

Intel S1200BTS rear I/O
Rear I/O ports for the Intel S1200BTS

 

Included Items

Typical of a server-class product, few items are included with this motherboard. For the most part, this is perfectly fine, although we never like to see a motherboard ship with less SATA cables than there are ports on the motherboard. In this case, there are 6 ports on the board, yet the motherboard only comes with two SATA cables.

The included items are:

  • 2x SATA Cables

  • I/O Shield

  • Quickstart Guide

  • Motherboard Schematic Sticker

  • Driver/Utility CD

Intel S1200BTS accessories

 

Motherboard Layout

Intel S1200BTS Intel S1200BTS diagram

Like most server motherboards, this board is somewhat limited in terms of the type and number of internal headers and ports. As we mentioned earlier, this is in fact a very good thing for a server motherboard as there are fewer potential points of failure.

Overall, we have few complaints over the layout of the components on the motherboard. Since this board is primarily designed for use in a rackmount chassis, the 24-pin ATX power connector (letter "L" in the diagram) is along the top of the motherboard rather than the right side. This is due to the fact that rackmounts are heavily dependent on unimpeded front-to-back airflow and moving the 24-pin power connector keeps it from blocking the airflow.

The only real "problem" we have with the layout is the location of the front switch/LED header (letter "X" in the diagram) which is located below the PCI slot. This is a very typical location for this header on Intel motherboards, but we much prefer to have the header closer to the front of the motherboard near the SATA ports. Plugging/unplugging cables from that header can be a chore at times, so the more exposed the header, the better.

RAID Functionality

Ever since Puget System began testing the onboard RAID functionality on motherboards, we have experienced consistant stability issues. Most of the problems boiled down to the controller not responding well to drives spinning-up at slightly different intervals. This becomes very apparent when we test onboard RAID controllers using a large number of standby loops. Platter hard drive will intermittently take slightly longer to spin-up and onboard RAID controllers simply do not handle this well. This usually results in a degraded array and considering that the RAID software is typically much more limited than a dedicated RAID card's software, it becomes a very big deal just to get the RAID back up and running. Taking into account that we have seen this happen after as few as 20 standby loops, this makes onboard RAID not very attractive when using platter hard drives.

Intel S1200BTS device manager

The Intel S1200BTS however has an actual LSI RAID controller built into the motherboard. The 6 internal SATA 3Gb/s ports can be switched in the BIOS between the normal Intel C202 controller and this LSI RAID controller. This controller does require a driver to be loaded during Windows setup, and since the DVD drive is also on that controller (since you cannot switch individual ports) this means that you have to load the driver from a USB key rather than the driver disk. This makes the Windows installation a bit more of a hassle since you need a second machine running in order to download the driver from Intel.com onto a USB key but we had no problems installing Windows once we had the driver loaded. 

While this is a big step in the right direction for onboard RAID, there are a few limitations. First, only a single RAID can be configured though the BIOS utility and any secondary RAIDs must be setup from within the OS with the RAID software. Second, since enabling the RAID is an all-or-nothing affair, even if you want to have a single disk as your primary, you will still need to configure it as a single disk RAID. We found that even a single disk RAID can become degraded and forced offline by power brown or black outs. This will cause the OS to crash and require a hard reboot. In order to bring the RAID back online so that the OS can load, you must load the RAID's BIOS configuration utility (Ctrl-E during POST), select the array and select the option "Force Online".

Intel RAID Web Console 2

This RAID controller will also only support RAID levels 0 and 1, as well as the combination of the two: RAID 10. There is very little configuration that can be done to the array (stripe size cannot be changed from 64KB for example) so you should not expect this RAID to be a top performer. For mirror (RAID1) arrays where speed is not an issue however, this should not be as much of a conscern.

As for the good parts of this RAID controller, there are two main points. First, the software for this card is also much better than the typical onboard RAID software and utilizes the Intel RAID Web Console 2. This is the same software that is used with the Intel RS2BL040 and RS2BL080 and has worked very well for us in the past. Second, the only time we experienced any stability problems on any type of RAID with both platter hard drives and SSDs was in the event of a power loss or brownout. Our standby loop testing - which has been very good at revealing RAID stability problems in the past - passed without any problems in all of our test configurations. This means that this onboard controller should be just as stable as an entry-level RAID card.

When we did manage to break an array, we received a pop-up alert that let us know that the RAID was degraded. From there, we were able to open the Intel RAID Web Console 2 and rebuild the array. A RAID1 array with two Western Digital Blue 500GB drives took roughly two hours to perform and a RAID1 with two Intel 510 250GB SSDs took roughly 45 minutes.

Intel RAID Web Console 2 Intel RAID Web Console 2
An alert pops up when a degraded array is detected The main page of the Intel RAID Web Console 2 shows an alert that there is a problem with an array
Intel RAID Web Console 2 Intel RAID Web Console 2
On the logical drive listing page, you can select the degreaded array and have it start the rebuilding process if it has not already begun Once the process completes, the error alert is removed

 

RAID Performance

We know that the LSI onboard RAID controller is up to par in terms of configuration and stability, so the remaining question now is performance. To test the performance of the onboard LSI controller, we used three different hard drives and compared the results to three dedicated RAID controllers. The full list of our test components is below:


The three different types of hard drives will give us a good look at the RAID performance across a broad spectrum of hard drive speeds. While both the Western Digital Caviar Blue 500Gb and the Intel 510 250GB SSD are actually SATA 6Gb/s drives, we only expect the Intel 510 250GB SSD to show any noticeable performance decrease while using the onboard RAID controller. The Western Digital drive is simply not fast enough to utilize the extra bandwidth available on a SATA 6Gb/s controller. Since the Intel 510 250GB will definitely max out the onboard controller, this will let us know the absolute performance limit of this controller.

The multiple RAID cards also cover a large span of performance options ranging from the entry-level LSI 9211-4i to the high-end Intel RS2BL080 and RS25DB080. Since the onboard RAID only supports RAID levels 0 and 1, we will only be looking at the results for those types of RAID. In order to get a feel for how well this this controller functions with just a single drive (if you wanted to have your secondary hard drives in RAID but not your primary drive) we will also be testing with one hard drive in a single disk RAID 0 array.

One thing to note is that the Intel RS25DB080 RAID card has 1GB of cache embedded which greatly enhances the random read and write performance.  Since this onboard controller does not have any caching, we will largely be ignoring the random read and write performance of that card since we do not expect it to come anywhere near the performance of the Intel RS25DB080.


Western Digital Caviar Blue 500GB

 

Single Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 125 MB/s 124 MB/s 125 MB/s 192 MB/s
Seq. Write 123 MB/s 121 MB/s 124 MB/s 123 MB/s
Rand. Read 512KB 44 MB/s 51 MB/s 75 MB/s 337 MB/s
Rand. Write 512KB 62 MB/s 61 MB/s 101 MB/s 217 MB/s
Rand. Read 4KB (QD=32) 1.0 MB/s
(252 IOPS)
1.2 MB/s
(304 IOPS)
1.3 MB/s
(313 IOPS)
1.8 MB/s
(436 IOPS)
Rand. Write 4KB (QD=32) 1.2 MB/s
(296 IOPS)
1.2 MB/s
(284 IOPS)
1.8 MB/s
(448 IOPS)
3.2 MB/s
(775 IOPS)

Starting with the slowest hard drive in a single disk configuration, we see little difference among our test controllers. The onboard RAID's random read is a bit lower than we would like to see, but it isn't too bad.

RAID 1 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 122 MB/s 64 MB/s 145 MB/s 168 MB/s
Seq. Write 94 MB/s 61 MB/s 123 MB/s 122 MB/s
Rand. Read 512KB 45 MB/s 25 MB/s 77 MB/s 394 MB/s
Rand. Write 512KB 54 MB/s 21 MB/s 99 MB/s 216 MB/s
Rand. Read 4KB (QD=32) 2.3 MB/s
(568 IOPS)
1.4 MB/s
(344 IOPS)
2.4 MB/s
(584 IOPS)
2.8 MB/s
(673 IOPS)
Rand. Write 4KB (QD=32) .95 MB/s
(231 IOPS)
.6 MB/s
(153 IOPS)
1.5 MB/s
(360 IOPS)
2.5 MB/s
(622 IOPS)

With the Western Digital Blue 500GB drives in a 2 disk RAID 1 array, we start to see some deviation between the different controllers. In this configuration, the performance of the onboard RAID is much better than the LSI 9211-4i, but not as good as the two Intel cards.

RAID 0 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 244 MB/s 168 MB/s 274 MB/s 355 MB/s
Seq. Write 238 MB/s 149 MB/s 243 MB/s 241 MB/s
Rand. Read 512KB 47 MB/s 47 MB/s 78 MB/s 363 MB/s
Rand. Write 512KB 85 MB/s 82 MB/s 143 MB/s 208 MB/s
Rand. Read 4KB (QD=32) 2.4 MB/s
(579 IOPS)
2.9 MB/s
(714 IOPS)
2.8 MB/s
(685 IOPS)
4.5 MB/s
(1096 IOPS)
Rand. Write 4KB (QD=32) 2.0 MB/s
(484 IOPS)
2.1 MB/s
(518 IOPS)
2.9 MB/s
(699 IOPS)
5.7 MB/s
(1383 IOPS)

In our final RAID configuration, we again see better sequential read and write performance than the LSI card (even almost matching the Intel RS2BL080!) but the random write 4KB with a QD of 32 is slightly slower than on the LSI card. Overall however, we can conclude that with the Western Digital Blue 500GB drive, the onboard RAID outperforms the LSI 9211-4i, but can't quite match the performance numbers of the two Intel cards.


Western Digital RE4 2.0TB

 

Single Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 145 MB/s 146 MB/s 158 MB/s 147 MB/s
Seq. Write 143 MB/s 143 MB/s 151 MB/s 144 MB/s
Rand. Read 512KB 55 MB/s 61 MB/s 91 MB/s 483 MB/s
Rand. Write 512KB 97 MB/s 99 MB/s 128 MB/s 258 MB/s
Rand. Read 4KB (QD=32) 1.8 MB/s
(437 IOPS)
2.0 MB/s
(499 IOPS)
2.1 MB/s
(507 IOPS)
2.8 MB/s
(692 IOPS)
Rand. Write 4KB (QD=32) 2.0 MB/s
(496 IOPS)
2.0 MB/s
(498 IOPS)
2.9 MB/s
(701 IOPS)
4.7 MB/s
(1147 IOPS)

Moving on to the Western Digital RE4 2.0TB drive, we again see fairly similar performance across the board in a single disk configuration. The onboard RAID does suffer slightly in the random read tests, but not by a large margin.

RAID 1 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 145 MB/s 74 MB/s 168 MB/s 183 MB/s
Seq. Write 118 MB/s 34 MB/s 144 MB/s 144 MB/s
Rand. Read 512KB 56 MB/s 29 MB/s 98 MB/s 667 MB/s
Rand. Write 512KB 83 MB/s 24 MB/s 126 MB/s 249 MB/s
Rand. Read 4KB (QD=32) 2.9 MB/s
(701 IOPS)
1.8 MB/s
(451 IOPS)
3.5 MB/s
(852 IOPS)
3.0 MB/s
(727 IOPS)
Rand. Write 4KB (QD=32) 1.9 MB/s
(462 IOPS)
.9 MB/s
(222 IOPS)
3.0 MB/s
(729 IOPS)
4.7 MB/s
(1146 IOPS)

In a RAID 1 configuration, we know from experience that the LSI 9211-4i has some performance issues. While the onboard RAID still does not match either of the Intel RAID cards, we are happy to see that it does not suffer from the same performance issues as the LSI 9211-4i.

RAID 0 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 296 MB/s 169 MB/s 289 MB/s 342 MB/s
Seq. Write 283 MB/s 161 MB/s 286 MB/s 286 MB/s
Rand. Read 512KB 60 MB/s 57 MB/s 94 MB/s 755 MB/s
Rand. Write 512KB 122 MB/s 109 MB/s 183 MB/s 248 MB/s
Rand. Read 4KB (QD=32) 2.9 MB/s
(715 IOPS)
3.7 MB/s
(911 IOPS)
3.7 MB/s
(902 IOPS)
4.8 MB/s
(1169 IOPS)
Rand. Write 4KB (QD=32) 3.9 MB/s
(956 IOPS)
3.9 MB/s
(958 IOPS)
5.6 MB/s
(1378 IOPS)
8.9 MB/s
(2181 IOPS)

In the RAID 0 configuration, the onboard RAID performs very well in our sequential read and write tests, matching even the Intel RS2BL080. In the random read and write tests we again see better performance than the LSI card, but it is still unable to match either of the Intel cards.


Intel 510 250GB SSD

 

Single Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 263 MB/s 323 MB/s 483 MB/s 598 MB/s
Seq. Write 247 MB/s 164 MB/s 321 MB/s 321 MB/s
Rand. Read 512KB 200 MB/s 265 MB/s 362 MB/s 902 MB/s
Rand. Write 512KB 231 MB/s 252 MB/s 380 MB/s 639 MB/s
Rand. Read 4KB (QD=32) 83.5 MB/s
(20381 IOPS)
80.5 MB/s
(19659 IOPS)
89.0 MB/s
(21724 IOPS)
194.5 MB/s
(47484 IOPS)
Rand. Write 4KB (QD=32) 50.5 MB/s
(12321 IOPS)
49.7 MB/s
(12140 IOPS)
50.6 MB/s
(12354 IOPS)
51.4 MB/s
(12550 IOPS)

On our final hard drive, we have to remember that neither the onboard nor the LSI 9211-4i support SATA 6Gb/s, so we expect to see much lower scores on those devices.  Interestingly, although the onboard has shown better single disk performance in all of our other configurations, the LSI 9211-4i actually has better sequential read and random read/write performance with this drive. The number of IOPS and sequential write scores is still better on the onboard however.

RAID 1 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 344 MB/s 169 MB/s 469 MB/s 736 MB/s
Seq. Write 233 MB/s 86 MB/s 324 MB/s 323 MB/s
Rand. Read 512KB 199 MB/s 135 MB/s 368 MB/s 896 MB/s
Rand. Write 512KB 209 MB/s 68 MB/s 378 MB/s 638 MB/s
Rand. Read 4KB (QD=32) 141.4 MB/s
(34529 IOPS)
81.3 MB/s
(19847 IOPS)
139.6 MB/s
(34083 IOPS)
207.3 MB/s
(50612 IOPS)
Rand. Write 4KB (QD=32) 49.8 MB/s
(12161 IOPS)
2.2 MB/s
(545 IOPS)
50.5 MB/s
(12325 IOPS)
51.4 MB/s
(12559 IOPS)
In a mirrored array, we again see much better scores with the onboard controller than with the LSI card. The two Intel cards are both leagues above the onboard and LSI scores, although that is to be expected from cards that support SATA 6Gb/s.

 

RAID 0 - 2 Disk Onboard RAID LSI 9211-4i Intel RS2BL080 Intel RS25DB080
Seq. Read 485 MB/s 182 MB/s 670 MB/s 980 MB/s
Seq. Write 74 MB/s 169 MB/s 649 MB/s 653 MB/s
Rand. Read 512KB 271 MB/s 232 MB/s 471 MB/s 995 MB/s
Rand. Write 512KB 70 MB/s 251 MB/s 610 MB/s 674 MB/s
Rand. Read 4KB (QD=32) 142.5 MB/s
(34787 IOPS)
158.3 MB/s
(38649 IOPS)
148.7 MB/s
(36314 IOPS)
366.3 MB/s
(89419 IOPS)
Rand. Write 4KB (QD=32) 6.8 MB/s
(1642 IOPS)
100.0 MB/s
(24405 IOPS)
100.8 MB/s
(24607 IOPS)
102.5 MB/s
(25023 IOPS)

In our final configuration we received some unexpected results with the write scores on the onboard RAID. Until now, the write scores have been right where we expected them, but for some reason the scores in this configuration suffered a significant performance hit on the onboard RAID controller.


Overall, we were very happy with the performance of the onboard LSI RAID controller. Except when in a RAID 0 with SSD hard drives, we saw overall better performance with the onboard RAID than with the LSI 9211-4i. Considering that you would only use an onboard controller when budget is a major concern however means that it is very unlikely that many users will be pairing high performance SSD drives with onboard RAID.

Sure, the onboard RAID did not get close to the performance of either Intel RAID cards, but you should remember the adage "you get what you paid for". Considering the onboard RAID is essentially free, the performance is more than satisfactory.

Conclusion

Overall, we found few problems with this motherboard. The feature set of this board is somewhat limited (which is to be expected on a server-class motherboard) so there were few things that could have caused us problems in the first place. The onboard LSI RAID controller worked out much better than we expected; having great configuration software (Intel Web Console 2) and RAID performance much higher than we expected. In fact, the onboard LSI RAID performed better than our current entry-level RAID card.

Speaking of the onboard RAID, there were a few problems with it. First, you can only configure a single RAID through the BIOS. Additional RAIDs will need to be setup using the software from within Windows. Second, Windows does not have native drive support for the onboard controller, so a driver must be loaded during setup. This is further complicated by the fact that the CD/DVD drive is on the same controller, so you must download the driver to a USB device and use that to load the driver. The configuration options are also very limited with RAID 1, 0 and 10 being the only RAID types available. You also cannot change many of the performance options like the stripe size from the default 64KB. Lastly, if you want to have a single drive not in RAID it can still suffer complications in the event of a power brown/black out since it has to be setup as a single disk RAID. Like a normal RAID, if this happens it must be forced back online through the BIOS configuration utility.

Intel S1200BTS Asus P8B-M
Intel S1200BTS Asus P8B-M

Most of the features we talked about in this article (including the onboard LSI RAID) are also available on the Asus P8B-M, plus the P8B-M has two SATA 6Gb/s ports and the option of installing a remote management module. So why would we consider the Intel S1200BTS over the Asus P8B-M? The answer is pretty simple: software certifications. Intel is very good about getting it's server motherboards certified for server-class software such as VMware. While the Asus P8B-M has the same chipset as the Intel S1200BTL, certifications only apply to a specific motherboard, not the overall chipset.

What it comes down to is if the advantage of having server-class software certifications is worth the cost of not having SATA 6Gb/s support and no remote management module options. Due to the popularity of the remote management modules on the Asus P8B-M, we have decided at this time to not move our product line to the Intel S1200BTS. There are PCI-based remote management cards on the market, but with a starting price of over $250 it simply is not worth the additional cost. We will continue to keep an eye out for a more cost effective remote management solution, and if one becomes available we will re-consider replacing the Asus P8B-M with the Intel S1200BTS.

Guest

Some of the results for the Intel 510 250GB SSD w/ the RS25DB080 seem a bit incredulous.

For example, with single disk, it pulls 598MB/s seq. read and 321MB/s seq. write speeds, but as per Intel:

http://ark.intel.com/products/...

The max is 500MB/s seq. read and 315MB/s seq. write.

Of course there's the 1GB cache, but even then if you write with file sizes larger than the size of the cache, the cache will have to write back to disk, and the disk should be the bottleneck for data rates.

What sizes were the files written during sequential r/w tests?

Posted on 2012-06-06 15:42:54