NDIII,

While the R430 does support SSD's, it doesn't look like the EVO 850's are going to be supported. Is there a reason you were wanting to use that specific drive?

Let me know.

Thank you for prompt reply Chris,

1.The EVO 850 was suggested to me by the dealer where I would like to  order the Dell server

Also I have read several good feedbacks about EVO 850 Series. and also I'm able to find it in dell store:  link

.But from another side I understand that there could be an issue with openmanage and also the RAID controller may show errors in logs that not acceptable to me as I need to be able fully control/view the ssd stats for in-time replacement like in production environment.

2. What about Samsung 845DC EVO ? Unlike EVO850 the DC EVO 850 is enterprise class SSD . Is it supported by H730 RAID controller ? by openmanage? Mb. I should proceed with DC version?

Thanks in advance.

Hello ,

1. Is it possible to use MegaCli tool with PERC H730 (as it LSI based controller) ? 

2.Could anyone confirm that it will be possible to view/monitor LifeTime/SMART stats (using openmanage , or any other way) with the next SSD's on the PERC H730 controller:

 a) Samsung 845 DC EVO (4xHW RAID-10)

 b) Samsung 845 DC Pro (4xHW RAID-10)

?

Any inputs are appreciated.

We're in the same boat - Dell partners and IT company in Miami, just purchased an R430 and clients needs changed (a few more VMs) and we need SSDs for performance. Have a PERC H730 card in there, running win svr 2012 r2.

Is there a list somewhere of major manufacturer SSDs supported? Perhaps the Samsung 850 PRO or intel Pro 2500 series?

Thanks

bump

Another vote for Samsung Pro 850 2TB SSD PERC 730 (and newer) official support.  We're planning on using it in a RAID-6 setup.  If not, Dell, there are other vendors supporting it already.

Did you try this out vastracts? I inherited a machine 12 of those 2TB drives on an H730 and it's been a massive headache. I just discovered a firmware update from November that i'm hoping resolves my issues (trying today), but I don't have high hopes.

Just got R430 with H730 mini. Decided to go with self-installed SSD, Dell SSD are ridiculously expensive.

Used two Intel 535 Series 240 GB SSD SSDSC2BW240H6R5 consumer grade in RAID 1 for OS (2012 R2) and three Intel DC S3610 Series 480GB SSDSC2BX480G401 data center grade in RAID 5 for SQL server data.

First of all, with all firmware/software up to date, there is no fault/warning lights on the front panel of R430. The only problem is that OMSA shows exclamation mark on these physical drives (status non-critical). Below info from one of my drives in OMSA.

ID 0:1:0

Status Non-Critical

Name Physical Disk 0:1:0

State Online

Power Status Not Applicable

Bus Protocol SATA

Media SSD

Revision RG21

T10 PI Capable No

Certified No

Capacity 223.00GB

Used RAID Disk Space 223.00GB

Available RAID Disk Space 0.00GB

Hot Spare No

Vendor ID

Product ID INTEL SSDSC2BW240H6

Serial No. CVTR5392076A240CGN

Part Number Not Available

Negotiated Speed 6.00 Gbps

Capable Speed 6.00 Gbps

Sector Size 512B

SAS Address 4433221104000000

The conclusion: no matter if HD is consumer grade or enterprise, if firmware in the drive is not from Dell, it will show non-critical in OMSA. It’s possible that other vendors may show even faults, I only tested with drives listed above. I’d suggest to get one drive and test first. There are some possible implications on performance. Will post my findings and performance results I get in another post.

I like to get the best performance possible, so I always test before going into production. All my tests done with AS SSD Benchmark 1.8.5636.37293, 3 GB load. Prefer it over Crystal Disk Mark as it shows IOPs in addition to speed.

What I found right away is that standard setup for RAID policy (Read Ahead [RA], Write Back [WB], Disk Cache Policy [DCP] disabled) gives me terrible write speed on RAID 1 with 535 series drives. After many tests I got best performance with RA, Write Through [WT], DCP enabled:

RAID size: 222.99 GB, Total Score – 996

Sequential: Read – 954.71 MB/s (59.67 iops); Write – 221.65 MB/s (13.85 iops)

Random 4K 1 thread: Read – 22.77 MB/s (5,829 iops); Write – 51.26 MB/s (13,123 iops)

Random 4K 64 threads: Read – 414.35 MB/s (106,073 iops); Write – 149.04 MB/s (38,155 iops)

Access time (512 B): Read – 0.147 ms (6,803 iops); Write – 0.788 ms (1,270 iops)

For comparison I created RAID 1 with DC 3610. With RA, WB and DCP disabled I got pretty good result:

RAID size: 446.62 GB, Total Score – 1051

Sequential: Read – 871.05 MB/s (54.44 iops); Write – 528.65 MB/s (33.04 iops)

Random 4K 1 thread: Read – 32.71 MB/s (8,375 iops); Write – 67.16 MB/s (17,192 iops)

Random 4K 64 threads: Read – 436.50 MB/s (111,743 iops); Write – 97.24 MB/s (24,893 iops)

Access time (512 B): Read – 0.086 ms (11,638 iops); Write – 0.169 ms (5,931 iops)

Unfortunately, I don’t have an explanation for it. Consumer grade vs data center? 535 series in particular? I guess at some point I’ll replace 535 with something from data center grade drives. For now, I’ll live with non-recommended settings for my OS RAID.

Now to data RAID 5. Here were no surprises, with default (and recommended) settings (RA, WB and DCP disabled) I’ve got the best results:

RAID size: 893.25 GB, Total Score – 1161

Sequential: Read – 1,649.04 MB/s (103.06 iops); Write – 1,038.38 MB/s (64.90 iops)

Random 4K 1 thread: Read – 32.35 MB/s (8,281 iops); Write – 55.80 MB/s (14,284 iops)

Random 4K 64 threads: Read – 428.57 MB/s (109,715 iops); Write – 76.76 MB/s (19,650 iops)

Access time (512 B): Read – 0.073 ms (13,733 iops); Write – 0.204 ms (4,906 iops)

To say that I’m completely pleased... no. RAID controller becomes a bottleneck for SSDs on writes in RAID 5. And understandable so, there are parity calculations involved.

I’ve set one DC 3610 in pass-through mode (in OMSA got to physical disks, choose drive, Available Tasks -> Convert to non-RAID disks) just to show how single disk performs. The results:

Disk size: 447.13 GB, Total Score – 950

Sequential: Read – 525.79 MB/s (32.86 iops); Write – 433.50 MB/s (27.09 iops)

Random 4K 1 thread: Read – 35.00 MB/s (8,961 iops); Write – 107.86 MB/s (27,613 iops)

Random 4K 64 threads: Read – 306.53 MB/s (78,471 iops); Write – 195.01 MB/s (49,923 iops)

Access time (512 B): Read – 0.032 ms (31,341 iops); Write – 0.097 ms (10,341 iops)

Tried to set 3 SSDs in pass-through mode and Windows Storage Spaces. Performance is much worse than hardware RAID on writes. Cannot even consider that.

My conclusions: test before you go into production. Get high endurance SSDs. If your load is 50/50 random reads/writes, RAID 1 will be fine. If random 70/30 (R/W), RAID 5. Three disks in RAID 5 are enough. Adding disks will most likely add insignificant read performance, but will reduce write performance (more parity to calculate). I wish I had enough disks to test RAID 6, 10, 50, but don’t think it will improve performance by much. On the other hand, it will add redundancy.

For my scenario, SQL server with mix of OLTP, OLAP and reporting services. I’m not an SQL wizard, but I expect 70/30 load, so RAID 5 will be fine for me.

Misc.: 535 has 5 years of warranty with endurance of 40 GB a day. I don’t expect more load for OS drive. DC S3610 has endurance of 3 full drive writes a day (3.7 PB in 5 years). Should be fine for busy SQL server.