February 1, 201214 yr Alright I have my HP N40L setup right now. I don't have a Parity drive(comes today) or cache drive in my system as of now. I also have read this http://lime-technology.com/wiki/index.php?title=Improving_unRAID_Performance. Unraid 4.7 HP Tweaks: Write caching turned on in the BIOS VGA set to 32MB in the BIOS PCI Power saving turned on in BIOS Hard Drive Speed Reads -/dev/sda: disk1 VB0250EAVER Timing cached reads: 2418 MB in 2.00 seconds = 1209.23 MB/sec Timing buffered disk reads: 292 MB in 3.01 seconds = 97.12 MB/sec root@HPTower:~# hdparm -tT /dev/sdb /dev/sdb:disk2 WDC_WD10EADS-00L Timing cached reads: 2430 MB in 2.00 seconds = 1215.59 MB/sec Timing buffered disk reads: 262 MB in 3.02 seconds = 86.76 MB/sec root@HPTower:~# hdparm -tT /dev/sdc /dev/sdc:disk3 WDC_WD10EADS-00L Timing cached reads: 2376 MB in 2.00 seconds = 1187.95 MB/sec Timing buffered disk reads: 264 MB in 3.01 seconds = 87.74 MB/sec root@HPTower:~# hdparm -tT /dev/sdd /dev/sdd:disk4 WDC_WD20EARS-00M Timing cached reads: 2372 MB in 2.00 seconds = 1185.86 MB/sec Timing buffered disk reads: 356 MB in 3.00 seconds = 118.53 MB/sec My setup (all devices gigabit) Netgear Router | | | | DlinkSwitch DlinkRouter | | | | Unraid Server PC1 Here is my ethtool eth0 Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Advertised auto-negotiation: Yes Speed: 1000Mb/s Duplex: Full Port: Twisted Pair PHYAD: 1 Transceiver: internal Auto-negotiation: on Supports Wake-on: g Wake-on: g Current message level: 0x000000ff (255) Link detected: yes Questions: If I have my Unraid server off of my Dlink Switch and my PC off of my Dlink Router I get like 11mb/s transfer speeds. If I have my Unraid Server on the same switch as my PC I get like 25mb/s transfer speeds. In my above example I put the Unraid server with the PC so on the Dlink Router. Do these speeds sound right? Why would it be slow on my original layout?
February 1, 201214 yr I have the same proliant server as you. I am no IT expert but i would hazzard a guess that there are more hops for it to go through hence the slower speed, i average transfer speeds of 40MB/s but i have a wired CAT6 network in my house with GB equipment, apparently the only let down in speed is the cache drive being 5400rpm blue WD where i might have slightly faster speeds if they were 7200 or even 10000rpm. Have you physically swapped round the places of the d-link and the switch ??, you seem to say you moved the unraid server and pc around but not tried changing the positions of the switch/routers, it might help eliminate other issues. Sorry i cant really be much more help than that, i am sure someone here will be able to ask you more complex questions
February 1, 201214 yr I likewise have the N40L micro server and the speed depends upon your drives and where on the drive it writes to. I am running Hitachi 3TB discs (5400 rpm) and writing directly to the protected array (No Cache drive) from my PC which is connected to the same switch I get 25 MB/s. From my QNAP NAS TS239 I get less then 20 MB/s also being connected to the same switch. For me I guess it is a combination of how fast the N40L can get served and then the Hardware of the N40L how fast can it write. I tried all kinds of different ways to improve the writing speed nut without any success and the common consensus here is that with Hitachi 3TB drives half full plus writing directly to the array 25 MB/s is about it. EDIT: Just turned on the Write Cache plus reduced VGA memory to 32MB as you did, which improved the write speed from QNAP NAS to N40L for 1.5 Gig file to 30 - 33 MB/s
February 1, 201214 yr Do you both have the hacked bios which unloops the ODD and Esata headers which then lets you write to those controllers at full speed ??..i also got an 8gb ram kit for this baby. £30.00, cant go wrong with that, i have just installed cache_dirs as well with /boot/cache_dirs -d 5 -m 3 -M 5 -w
February 1, 201214 yr Author Yes on the hack part. dmesg | grep SATA | grep link ata5: SATA link down (SStatus 0 SControl 300) ata6: SATA link down (SStatus 0 SControl 300) ata4: SATA link up 3.0 Gbps (SStatus 123 SControl 300) ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300) ata3: SATA link up 3.0 Gbps (SStatus 123 SControl 300) ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
February 1, 201214 yr Author I likewise have the N40L micro server and the speed depends upon your drives and where on the drive it writes to. I am running Hitachi 3TB discs (5400 rpm) and writing directly to the protected array (No Cache drive) from my PC which is connected to the same switch I get 25 MB/s. From my QNAP NAS TS239 I get less then 20 MB/s also being connected to the same switch. For me I guess it is a combination of how fast the N40L can get served and then the Hardware of the N40L how fast can it write. I tried all kinds of different ways to improve the writing speed nut without any success and the common consensus here is that with Hitachi 3TB drives half full plus writing directly to the array 25 MB/s is about it. EDIT: Just turned on the Write Cache plus reduced VGA memory to 32MB as you did, which improved the write speed from QNAP NAS to N40L for 1.5 Gig file to 30 - 33 MB/s Are you using Jumbo frames at all - I am not for my test purposes.
February 2, 201214 yr I do not use the hack. I only run 4 drives from the built in controller that has the full throughput enabled. I don't use the ODD & eSata connections. I have a rocket Highpoint 620 two port controller in one of the slots for the additional two drives should the prices for 3TB drives ever come down in prices to a reasonable level. In the second slot I have an Intel NIC that is excellent in performance which also allows jumbo frames and all my equipment, QNAP NAS, 2nd Unraid server, PC1, PC2, Switch can all handle jumbo frames. So I am running my N40L with jumbo frames = 9000. The stock N40L has an onboard NIC that does not allow for jumbo frames, you have to have a separate NIC. My N40L has 4Gig as memory and I am running unMenu as well as the cache_dirs script without any problems. So far I am very happy with the behavior of version 5b14 with my N40L and with the understanding that with the given hardware (processor & disc drives) I probably have reached the max in writing speed. It is not a huge problem for me since the writing of files to the server are only TV episodes and that takes place every other day and takes a few minutes. Running parity check will take 36 to 40 hours with 3TB drives but that is maybe every other month, so all is good and doable.
February 2, 201214 yr Author I do not use the hack. I only run 4 drives from the built in controller that has the full throughput enabled. I don't use the ODD & eSata connections. I have a rocket Highpoint 620 two port controller in one of the slots for the additional two drives should the prices for 3TB drives ever come down in prices to a reasonable level. In the second slot I have an Intel NIC that is excellent in performance which also allows jumbo frames and all my equipment, QNAP NAS, 2nd Unraid server, PC1, PC2, Switch can all handle jumbo frames. So I am running my N40L with jumbo frames = 9000. The stock N40L has an onboard NIC that does not allow for jumbo frames, you have to have a separate NIC. My N40L has 4Gig as memory and I am running unMenu as well as the cache_dirs script without any problems. So far I am very happy with the behavior of version 5b14 with my N40L and with the understanding that with the given hardware (processor & disc drives) I probably have reached the max in writing speed. It is not a huge problem for me since the writing of files to the server are only TV episodes and that takes place every other day and takes a few minutes. Running parity check will take 36 to 40 hours with 3TB drives but that is maybe every other month, so all is good and doable. What do you mean by Full Throughput enabled? Also I didn't realize the N40L didn't support Jumbo Frames. I have some spare gigabit nics that do that I can test out.
February 2, 201214 yr Sorry, not enabled as in enabling a setting, but full through as in the max performance that the stock N40L will give to the 4 data drives in comparison to the throttled speed to the ODD. No special mods for BIOS or any other tweaks, just stock setup. I also was shocked that the N40L has such outdated NIC chipset that will not allow for jumbo frames. Search the comments on Newegg for jumbo frames and you will see that others are disgusted also. Search the forum here for jumbo frames and again you will find some excellent comments and I believe in the end jumbo frames should not make a huge difference since we are not out on the Net where it can make a huge difference but on our own home network.
February 2, 201214 yr Apparently, Jumbo frames is not such a useful feature in 2012 -- it was mainly to help save CPU cycles in the past. On my HP N36L (same machine with a lower speed CPU and FSB), I can get up to 40MB/sec to a 5400rpm cache drive. It could be related to the machine you are sending from? Try using midnight commander to copy data from one drive to another to check write speeds without using the LAN.
February 2, 201214 yr Even try FTP, I have N36L as well, and I could reach 100MB/s to a 7200 cache drive
February 2, 201214 yr Author I did a test with MC from Disk 4 to Disk2 - It topped out at like 46mb/s does that sound right? /dev/sdd:disk4 WDC_WD20EARS-00M Timing cached reads: 2372 MB in 2.00 seconds = 1185.86 MB/sec Timing buffered disk reads: 356 MB in 3.00 seconds = 118.53 MB/sec /dev/sdb:disk2 WDC_WD10EADS-00L Timing cached reads: 2430 MB in 2.00 seconds = 1215.59 MB/sec Timing buffered disk reads: 262 MB in 3.02 seconds = 86.76 MB/sec
February 2, 201214 yr Similar values here, but then I do NOT write to a cache drive but directly to the array which accounts for the lower speeds as I don't really see the need for sticking a 2.5" drive in there to gain a few minutes of time. The peace of mind knowing that it is always protected is worth it. And I agree that the machine I am sending from does make quite a bit of difference in performance. There is a significant speed increase when transferring from my QuadCore i7 950, 3200 MHz (24 x 133) with 12 Gig of memory versus my QNAP NAS with an Intel Atom Processor 1.6 GHz with 1GB DDRII RAM plus running 2 WD 1.5TB discs in Raid 1 (mirror) configuration. Especially is it noticeable when the NAS is doing other things, like Unraring, Downloading and what have you. And as already mentioned, I did not perceive any real apparent speed increase using jumbo frames, having spent $22 on a NIC it stays there until I need the slot.
February 2, 201214 yr i think theres more benefit to the cache drive than just a few minutes, like cache_dirs...though if that would work on the array i might be tempted to dump the cache drive..if i took apart the N40l to get full access to the top of that cage i reckon i could fit in a 3.5" 2TB instead of a laptop drive, height wise it fits, you just cant get an obtuse enough angle to slide it under
February 2, 201214 yr I did a test with MC from Disk 4 to Disk2 - It topped out at like 46mb/s does that sound right? If parity is not assigned, yes.
February 3, 201214 yr Alright I have my HP N40L setup right now. I don't have a Parity drive(comes today) or cache drive in my system as of now. I also have read this http://lime-technology.com/wiki/index.php?title=Improving_unRAID_Performance. Unraid 4.7 HP Tweaks: Write caching turned on in the BIOS VGA set to 32MB in the BIOS PCI Power saving turned on in BIOS Hard Drive Speed Reads -/dev/sda: disk1 VB0250EAVER Timing cached reads: 2418 MB in 2.00 seconds = 1209.23 MB/sec Timing buffered disk reads: 292 MB in 3.01 seconds = 97.12 MB/sec root@HPTower:~# hdparm -tT /dev/sdb /dev/sdb:disk2 WDC_WD10EADS-00L Timing cached reads: 2430 MB in 2.00 seconds = 1215.59 MB/sec Timing buffered disk reads: 262 MB in 3.02 seconds = 86.76 MB/sec root@HPTower:~# hdparm -tT /dev/sdc /dev/sdc:disk3 WDC_WD10EADS-00L Timing cached reads: 2376 MB in 2.00 seconds = 1187.95 MB/sec Timing buffered disk reads: 264 MB in 3.01 seconds = 87.74 MB/sec root@HPTower:~# hdparm -tT /dev/sdd /dev/sdd:disk4 WDC_WD20EARS-00M Timing cached reads: 2372 MB in 2.00 seconds = 1185.86 MB/sec Timing buffered disk reads: 356 MB in 3.00 seconds = 118.53 MB/sec My setup (all devices gigabit) Netgear Router | | | | DlinkSwitch DlinkRouter | | | | Unraid Server PC1 Here is my ethtool eth0 Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Advertised auto-negotiation: Yes Speed: 1000Mb/s Duplex: Full Port: Twisted Pair PHYAD: 1 Transceiver: internal Auto-negotiation: on Supports Wake-on: g Wake-on: g Current message level: 0x000000ff (255) Link detected: yes Questions: If I have my Unraid server off of my Dlink Switch and my PC off of my Dlink Router I get like 11mb/s transfer speeds. If I have my Unraid Server on the same switch as my PC I get like 25mb/s transfer speeds. In my above example I put the Unraid server with the PC so on the Dlink Router. Do these speeds sound right? Why would it be slow on my original layout? 11MBps is correct if one or more of you network devices is connecting at 100Mbps. unRAID is connecting to the DlinkSwitch at 1000Mbps. What are to connection speeds to the Netgear Router, the DlinkRouter, and the PC?
February 3, 201214 yr My PC1 is set to autonegotiate. Check what speed is being negotiated. The other devices also need to be checked. Connect to the routers and confirm their connection speeds.
February 4, 201214 yr Author The dlink is set to auto but can't find anyplace on the Netgear - Interesting note is I found that the connection from the netgear to the Dlink switch is only at 100/Full. Not sure why - The above is routed the following way - It comes off the Netgear goes into the wall in the garage which then connects to the spare room opposite side of garage which then goes into a gigabit switch. Ok so I opened up the jack and it looks like the connection going to the front of the house was setup for a loopback as it is only running gr/wh br/wh so this would explain the 100/full connection.
February 6, 201214 yr i would 'guess' so..i am no IT expert but i believe that all the cabling on your network should be setup up the same, unless your some crazy guru type that needs something different for whatever reason, i wired my house myself and just googled the wiring diagrams, it works... http://www.structuredhomewiring.com/Images/WiringCat5OutletA.gif
Archived
This topic is now archived and is closed to further replies.