Jump to content
  • 0

budget media server build


duelistjp

Question

Recently got my first job and decided that even before a car I wanted to finally build myself a media server.  As it stands I'm using an ancient pc without even sata ports and it runs lubuntu with unimaginable lag.  Would probably be better of dual purposing one of the several modern desktops we have in the home but parents refuse to support a media or file server for some reason even though they have been very generous overall in terms of computers.  I'm still pretty poor though and want to get a car before too long so I'm on a bit of a budget.  talked my parents into getting fiberoptic in a few months when it is in our area so should have 200mbps down 50 mbps up.  There may be a few program isos and what not in addition to the media but no irreplaceable data on here so to keep cost down i won't have redundancy or backup of the data drives at least at first.  Will just output tree to a txt file once a week and save it on google drive.  It will run Plex, sickrage, couchpotato, transmission, and samba.  The most it should ever deal with is 3 streams rarely, normally 2 or less and generally one of the streams will be playing to a desktop on the local network so it shouldn't need transcoding on that.  Was hoping to get your opinions on the hardware i'm looking at.  Future expandability especially in storage size is important to me but I don't want to add any cards to the mobo to get it for some time. Here are the components i'm looking at although these are not necessarily where i will buy them and may have to make substitutions if price changes before I get them.  What I am most unsure about is the processor. am i likely to need an i5 for the load i'm planning on putting on it in plex.

 

1. case - http://smile.amazon.com/gp/product/B0091IZ1ZG/ref=oh_aui_detailpage_o09_s00?ie=UTF8&psc=1

 

2. mobo - http://www.newegg.com/Product/Product.aspx?Item=N82E16813157500

 

3. ram - http://www.newegg.com/Product/Product.aspx?Item=N82E16820231666

 

4. processor - http://www.newegg.com/Product/Product.aspx?Item=N82E16819117447

 

5. system drive - http://smile.amazon.com/gp/product/B00KFAGCWK/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1

 

6. data drives (will need more later) 2x - http://smile.amazon.com/gp/product/B00EHBERSE/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1

 

7. psu - http://www.newegg.com/Product/Product.aspx?Item=N82E16817139028

Link to comment
Share on other sites

Recommended Posts

  • 0

For Scanner, I'm not 100% sure. Besides ... disks haven't used CHS for addressing in forever, so it's harder to tell.

We start from LBA sector 0 and go up from there, IIRC. So yes, it should be in the "center" of the platter, theoretically.

 

As for scanning speed, it should all be sequential. But if the disk is doing something else, it will affect the scan speed.

And it could be the disk doing something internally, as well.

 

And i'm not surprised that it runs warmer.... makes sense.

 

 

As for the seek error rate, wikipedia says it best:

(Vendor specific raw value.) Rate of seek errors of the magnetic heads. If there is a partial failure in the mechanical positioning system, then seek errors will arise. Such a failure may be due to numerous factors, such as damage to a servo, or thermal widening of the hard disk. The raw value has different structure for different vendors and is often not meaningful as a decimal number

 

 

Basically, the drive isn't in our database yet, so we display raw values. 
Submit it to bitflock, and let me know the ID number. Then we can see about adding the correct SMART interpretation. 

 

 

 

 

 

As for the backups, if this is server backup, especially, the inital one, this doesn't surprise me at all.

The Server Backup feature uses VHDs and writes block by block, meaning that you should get at least decent performance out of it.

 

The real question is how fast it runs in a week, or month.

Link to comment
Share on other sites

  • 0

Yeah, but there is, in my situation, not a lot to write again. A few files and client backups, that is all that changes. Last night the backup finished in 22 minutes but I do not know how much it actually wrote.

If this was the server backup, then run "wbadmin.msc". Open an entry, and it will tell you how much data it transferred. :)

 

As for client backups... check the modified date? 

Link to comment
Share on other sites

  • 0

Never been there before, thanks! 30 mins for 41GB, that is 22.8MB/s. Of course, this is including preparing/comparing/finalising so a bad benchmark. To be sure, an older backup wrote 25GB in 37 mins but the target was a ST2000DM001 (Seagate Barracuda 7200?).

 

Edit: So I am looking at Scanner & Resource Monitor while a Server Backup is being made and, grab something to hold on to, I get 'high' scores like between 130 and 460KB/s. Not MB, KB! Response time, 3,621ms! Disk Queue runs as high as 87.16. Oh wait, all of a sudden numbers start to improve, 3MB/s, 170ms, rising as I write to 5, 6, 8, 12MB/s. I'm fine with that as long as backups succeed and take less than 12 hours (before the next one starts) but oh boy, does it seem SMR can BITE! Now it's at 30MB/s.

 

Anyway, it finished in 1:22 hrs, wrote 127.83GB according to wbadmin.msc, 26MB/s average. That's _write_, total I/O is higher as backup reads a lot to. A better benchmark/stress test I know not how to do (aside from running some SQL Server DB on it). Once my WD Red is up for Server Backup duty I'll have a look at that one too.

Link to comment
Share on other sites

  • 0

Glad I could show you that. It's also the go to place for diagnosing problems.

 

As for the speeds, that ... unfortunately seems about right. :(

There is a lot of reading here, because the backups ARE incremental. And IIRC, they're actually differentiated VHDx files. So it may not be an efficient storage solution for the SMR tech.

 

 

As for benchmarks, synthetic ones like ATTO and the like are common. But they're synthetic.

 

Copying files and actually using it are the best benchmarks, IMO. 

 

And again, yeah the SMR is going to bite for any sort of random access, which will happen for frequent access. 

Link to comment
Share on other sites

  • 0

WHS2011 so .vhd files, not vhdx unfortunately. I'll check for a while but overall. In two weeks my WD Red Backup HDD is up in the roster, I'll take a look at that as well and compare. So far, I am happy. They may not be the best write-performers due to SMR (still unsure) but they get the backup job done it seems, I can grow my server to about 8TB in unduplicated backupable data at low cost and they read like crazy.

 

It is a ST8000AS0002-1NA17Z and the ID number is J1PO37PU

Link to comment
Share on other sites

  • 0

Thanks for submitting. I've flagged alex, and we'll see about adding a proper interpretation for these drives.

 

 

And I meant to use "VHD(x)".

But either way, I'm glad to hear that it is working well.

And yea, the write performance is most likely due to the SMR technology (shingled magnetic recording).

Ironically, the best analogy is ... shingling a roof. It's easy when you do it layer by layer. But if you have to replace a single tile/shingle, it's a PITA to do so. 

Link to comment
Share on other sites

  • 0

I should probably stress it with some of my SQL stuff but I don't like to spend the time right now. On my lappy I have an SSD for the DBs but I had the tempdb on the spinner. When I moved it to the SSD as well it boosted a script by 40%. Hate to think what would happen on the Seagate Archive HDDs, I imagine queue depths well in the hundreds.

 

Anyway, love the wbadmin.msc, it there something similar for client backups in WHS2011?

Link to comment
Share on other sites

  • 0

Well, mine should arrive soon, and I will definitely be testing it out. :)

 

Anyway, love the wbadmin.msc, it there something similar for client backups in WHS2011?

I wish!

 

I'm not sure about WHS2011, but 2012R2 runs the following on clients:

 

 

C:\WINDOWS\System32\Essentials\RunTask.exe /asm:"C:\WINDOWS\System32\Essentials\BackupClientProvider.dll" /class:Microsoft.WindowsServerSolutions.DataProtection.PCBackup.ObjectModel.PCBackupClientManager /method:DoScheduledBackup /task:"Client Computer Backup"
.

 

The paths may be different for WHS2011. but you should find this in the Task Scheduler, under Microsoft\Windows Server (or similar).

Link to comment
Share on other sites

  • 0

Got my Seagate Archive 8TB drive! :)

 

Aside from the atrocious packing done by the seller...

https://drashna.net/owncloud/index.php/s/mZTNoUyWwUvZOnI

 

 

Well, doing some testing with synthetic benchmarks such as ATTO, the drive looks ... well, pretty on par with the Seagate NAS and WD Red drives in terms of performance.

 

Doing copies of large files (ISOs), I was getting around 120/s sustained write speeds to the disk (from my pool, also using 64k allocation unite size for the disk, which seems to make a 20MB/s difference in write speed).

File reads are around 140MB/s consistently.

 

In both cases, it does fluctuate a lot. And occasionally, writes would stop (0 bytes/s), but not often. 

 

 

All in all, I would say that the Seagate Archive drives should work well with DrivePool. Though, for best effect, the Ordered File Placement Balancer may get you the best results from these disks.

Link to comment
Share on other sites

  • 0

Oh yes I agree, as long as it is writing new tracks. But at some stage, like I see with Server Backups, writes do become slow at times. What I had hoped and turned out to be correct is "The drive leverages an on-drive cache (roughly 20GB) to handle inbound writes, in addition to internal systems for meta data tables and background processes like garbage collection, not unlike an SSD."

Link to comment
Share on other sites

  • 0

Yeah, I was reading that, and that part stuck out as well.

 

While writing, data to it, I noticed that periodically it would completely stall (0 bytes/sec) for a few seconds. 

 

Aside from that, the article indicated that reading was a priority in most cases, which is fantastic if you want to pool the drive. 

I've done so, and it works VERY well in the pool from what I'm seeing.

 

 

The SMR technology being used is more beneficial for read, or data retrieval, purposes like active archive

If this is actually the case, then it really means it's great for large pools that people like myself have. I keep the data for "archival" purposes (at least that's what I tell myself and my checkbook), because it can be hard to find stuff later on.

 

 

Instead we built out a Veeam backup test to create a similar data model. In our test we saw, as expected, the SMR drives took much longer for a traditional full backup, averaging 30MB/s. However we saw sustained read speeds during a 400GB VM recovery in excess of 180MB/s, which is really the core metric.

I'll second that "in excess of 180MB/s".

I was getting the best result with 64k cluster sizes (allocation unit size), though I suspect I would get better if I used powershell to go to 128k or 256k clusters instead... but that was just too much disk space wasted for me (lots of images).

But I was seeing 150-180MB/s consistently, and it peaked into the high 200's a few times (for a fairly "long" time" too).

 

 

As for the performance issues... coupled with read striping, I these drives would make a fantastic pool.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...