BUILD Good deal? SYS-5037C-I

Status
Not open for further replies.

rockstar0215

Dabbler
Joined
Dec 26, 2013
Messages
14
I've been lurking and basically had a build down. I planned on using the newest Haswell i3 processor on an X10 board with matching compatible RAM. However, just today a deal passed across my desk and I think I am going to pull the trigger. For about $120 someone is selling the SYS-5037C-I new in box. This system has a X9SCL board. I'd go with an Intel Xeon 1230v2 or a 1220v2. Both support ECC support. I would have gone with an i3 in the X10 system, but the non-haswell i3 does not support ECC. Big no no.

All in all, is this worth it? The entire thing comes with a power supply, a chassis, and an X9 board. I wanted the i3 because I wanted to save money, but it seems like I will save money with this regardless if I have to spend double on a Xeon processor.

The X9 is still good for freenas?

Thoughts on buying older hardware but at a significant discount?

I am excluding RAM for now because I have to do some research (aka read more of the sticky and do a general search)
 
Last edited:

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
Ivy Bridge is still perfectly acceptable, particularly if pricing is favorable (it usually isn't, as few systems have been replaced and the new stuff doesn't get aggressively marked-down).
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
How about the 300watt power supply. Is that sufficient to run a Xeon and 6 drives?

That's cutting it close. I'd definitely look at something beefier.
 

marbus90

Guru
Joined
Aug 2, 2014
Messages
818
The included PSU has 2x 12V rails at 12+16A with a cap at 216W - I don't like that setup. Single rail PSUs are better.
 

rockstar0215

Dabbler
Joined
Dec 26, 2013
Messages
14
Something like a Seasonic SSR360 OR SSR550?

I plan an array of 6 discs first and maybe add another 6 disks if need be. The 55o has 6 sata connectors.
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
Ivy Bridge is still perfectly acceptable, particularly if pricing is favorable (it usually isn't, as few systems have been replaced and the new stuff doesn't get aggressively marked-down).

Even Sandy is fine. The differential between Sandy and Ivy is modest.

How about the 300watt power supply. Is that sufficient to run a Xeon and 6 drives?

Sufficient to RUN them? Yes. Not sufficient to START them without risk of voltage droop, though, and that's what matters.

Something like a Seasonic SSR360 OR SSR550?

I plan an array of 6 discs first and maybe add another 6 disks if need be. The 55o has 6 sata connectors.

Generally speaking, a 360W supply is cutting it close for six disks.

The users here mostly really like the Seasonic G's.

https://forums.freenas.org/index.php?threads/i-need-some-data-points.32829/#post-204099

etc. show you my method of estimation for power supply sizing. The G-550 would probably come out as the ideal candidate if addition of six more drives later is a consideration. Otherwise, I'd suggest G-450.
 

mjws00

Guru
Joined
Jul 25, 2014
Messages
798
How many times does 6 go into 45? Is it 7.5? 50A / 7.5 = 6.66 = the beast = kill it with fire. Dang it where is my Devil emoji. ;)
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681

You're saying ... what? Designing storage systems at large scale is quite a different thing. We've been doing it here for many years.

I don't think that says anything that invalidates what I've been saying, in any case. Get a properly sized power supply. If you have the capabilities in house to do things like monitoring actual power consumption per rail, then that's very helpful. Most forum users here do not have that sort of gear, or even a proper electronics shop in which to build servers. A few of us are the exception to that rule.
 

marbus90

Guru
Joined
Aug 2, 2014
Messages
818
Did you read that link? It's about the energy in a Storagepod with 45 Desktop HDDs. There is said that 51A@12V is the max, ever measured, the highest, the biggest value for 45 HDDs spinning up. We don't have 45 HDDs here. We have 6 HDDs. mjws00 did the math for you. A 360W has megatons of overhead already. And a socket 1155 Xeon doesn't use much as well. Primary side usage is below 8A@12V.

My only problem: I do not trust split-rail PSUs. Got 3 at home, neither of them can feed many peripherals. One Seasonic S12II 330W single-rail however doesn't have problems, even when I'm throwing an X6 1090T and a HD4850 at it.
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
For an aggregate of 45 drives, not all of which are necessarily spiking at the same exact moment. Again, though, this still doesn't validate buying a supply that isn't sufficiently large to be able to provide the oomph to run all those drives simultaneously. The math says that there's a small cost in terms of PSU efficiency, probably a watt or two, in going with a slightly larger supply. The slightly larger supply might be a slightly higher cost.

It is unclear why you think building something that might not be powerful enough is a good idea, and I'm going to be dinging you more warning points if you continue to spam threads trying to further this pointless "debate."
 

malcolmputer

Explorer
Joined
Oct 28, 2013
Messages
55
For an aggregate of 45 drives, not all of which are necessarily spiking at the same exact moment."

More importantly, that link's analysis didn't even measure the voltage at that current spike. Since motor spin up is a constant power load, it is very possible that the 50A draw corresponded to a voltage sag, and since they didn't measure it who knows how bad it could have been. Furthermore, with 50A they would also have to measure at the drives to make sure they accounted for the voltage drop in the lines.
 

marbus90

Guru
Joined
Aug 2, 2014
Messages
818
If a PSU rated for 96A at 12V reacts to a 50A load with a significant voltage sag, it's defective. Note that ATX specification allows for +-10% voltage differences. Even if it would sag by 0.2, 0.5 or even 1V, it would be no big deal. It is inside the specifications.
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
With a sufficiently large supply (and theirs seems to qualify), I would not expect to see a voltage sag, though some flutter is possible.
 

malcolmputer

Explorer
Joined
Oct 28, 2013
Messages
55
With a sufficiently large supply (and theirs seems to qualify), I would not expect to see a voltage sag, though some flutter is possible.

Still something they should be measuring if they plan on publishing the data for other people to be using.
 

mjws00

Guru
Joined
Jul 25, 2014
Messages
798
Lots of reviews show sag and ripple at full load. That is easy, common, and frankly every decent modern supply has no trouble holding their advertised spec.

Putting up a beautiful graph of startup loads from a large system using a proper scope is AWESOME. This is NOT as easy. Few reviewers have a 10K+ fully populated 45 drive box sitting around with clamp meters they can hook to a scope.

Some things I love about that link:
  1. It looks EXACTLY as we expected it to. Every set of brushless motors spinning up with a constant load looks about like this. This is a nicely controlled ramp and reasonably gentle.
  2. It is NOT an estimate based on a datasheet or numbers pulled out of the air.
  3. The datasheets seem to be conservative. We are showing a Max current of 51A. 51A/45 drives is 1.13A per drive startup load. For ~4.5 seconds.
  4. Spun up load is (14A /45 ) 0.31A per drive spinning.
  5. It takes measuring at the mains and sampling on a kill-a-watt or UPS out of the equation. However, it also shows those numbers are pretty darn close.
** 12V Rail only. The 5V rail is not really that interesting.

Things to NOTE:
  1. They have a pretty cool active/active supply. Which is efficient at low loads with cool redundancy. Spec 950w. Designed for 1500w surge.
  2. ~Startup load 50A*12V = 600W, 19A*5V = 95W. So 695w total. Not unreasonable to think (695w/45d) 15.4w per drive startup.
  3. 695w/ 950w = 73.1% utilization of 2/3 of supply with 255w available + 475 on n+1 supply (1425w). So (695w/1425w) 48.7% of designed capacity. With surge to 1500w in datasheet.
  4. This is a very well designed (safe) system with plenty of excess capacity for hotter drives, additional cooling, compute etc.
It's wonderful data any way you slice it. This is just the numbers. Whether it makes sense to optimize for cost, and pick a quality supply with slightly less excess overhead is a different story. For me it is easy to justify modular and larger. Good PSU's often last me for many generations of upgrades and servers.
 

marbus90

Guru
Joined
Aug 2, 2014
Messages
818
"Standard Power Supply" Zippy PSL-6850M 850W -> $525 - for a nearly 10 year old PSU design. It isn't 80+ certified, it's only got 60A@12V.
"Redundant Power Supply" Zippy M3W-P6950L -> 1085.27$ + 60$ in chassis modifications, same old design, same low 12V ratings, also not 80+ certified.
That's quite a big asking price for a PSU, especially if you're going to use it at home and would have no use for Dothraki PSUs.

I've found that the Seasonic 850W and 1050W PSUs are pretty much the same price wise. While the 850W delivers 70A on 12V and 25A on 5V, the 1050W offers 87A on 12V. That ought to be plentiful enough for 45 drives, Xeon E5 (even though it won't be needed with Skylakes RDIMM support) and the HBAs. Even the 1050W one fits my 20% constant load sizing guide, so why not. It's a modern PSU which gets loaded to the same amount in gaming computers. And this is for hours on end, not just a small HDD spinup burst.

Oh, also note these are 7200rpm HDDs, not the more common 5400ish NAS HDDs. I'm sure you can save something there design-wise. Also they got staggered spinup working with the Rocket 750 card: http://www.45drives.com/wiki/index.php/Using_Staggered_Spin_Up
 
Status
Not open for further replies.
Top