manuti Posted January 19, 2017 Posted January 19, 2017 Orange Pi Zero NAS Expansion Board with SATA, mSATA using internal HUB USB with UASP capable chip, 2x USB and AV Portmore info in this cnx-software post: http://www.cnx-software.com/2017/01/19/orange-pi-zero-nas-expansion-board-with-sata-usb-and-av-port-sells-for-10-shipped/ I want to share this new expansion board to read some interesting feedback here.
manuti Posted January 19, 2017 Author Posted January 19, 2017 I fail to spot an USB hub there. I add two images from the other side of the board. I don't know if the JMICRON chip is an USB HUB or only the SATA and mSATA converter.
tkaiser Posted January 19, 2017 Posted January 19, 2017 I add two images from the other side of the board. I don't know if the JMICRON chip is an USB HUB or only the SATA and mSATA converter. The JMS578 is only an USB-to-SATA bridge making use of some SPI flash and containing voltage regulators (1.2V and 3.3V -- the latter necessary for mSATA). But there's no hub. Steven mentioned USB would be multiplexed (so maybe if you use only a mSATA SSD you get USB on the left USB jack and when using only the SATA port the right USB jack exposes the respective signals?) BTW: There's is also some information/speculation in the other thread: https://forum.armbian.com/index.php/topic/2808-orange-pi-zero-went-to-the-market/?p=23532
manuti Posted January 19, 2017 Author Posted January 19, 2017 The JMS578 is only an USB-to-SATA bridge making use of some SPI flash and containing voltage regulators (1.2V and 3.3V -- the latter necessary for mSATA). But there's no hub. Steven mentioned USB would be multiplexed (so maybe if you use only a mSATA SSD you get USB on the left USB jack and when using only the SATA port the right USB jack exposes the respective signals?) BTW: There's is also some information/speculation in the other thread: https://forum.armbian.com/index.php/topic/2808-orange-pi-zero-went-to-the-market/?p=23532 OK, thanks. You discover to me in other post the boards using USB to SATA converter and the bad decision of buying this cheat way of having SATA connections.
jimandroidpc Posted January 19, 2017 Posted January 19, 2017 Interesting idea, but not all that excited by the Fast Ethernet 10/100 only for NAS type applications. I think the minimum at this point should be 1000mbps for a NAS type device. For me - I dont think Id go to the trouble and $$ of putting an mSATA on there...
jimandroidpc Posted January 23, 2017 Posted January 23, 2017 I did some benchmarking with my Orange Pi Zero connected to my gigabit router. To my Windows Server PC I can get transfers around 85-100MB/ Sec, Banana Pi server ~30-35MB/sec. Same 1.3gb file to the Orange Pi Zero - 5MB/Sec. Pretty disappointing for a NAS device..Still will probably use it to run an offsite syncthing node for a few gb of family pics.
tkaiser Posted January 23, 2017 Posted January 23, 2017 Why benchmarks with a Fast Ethernet only device? Everything is bottlenecked by slow network (and cpufreq governor in this case -- switching to performance should give a few MB/s more). Here are numbers what to expect with an external USB3 GbE dongle: https://forum.armbian.com/index.php/topic/1440-h3-devices-as-nas/ but I really hope we see new Zero boards with Gigabit Ethernet soon (and maybe H5 instead of H3) and I also hope that the people who currently order this NAS Expansion board are aware that it makes nearly no sense combined with actual OPi Zero (especially not for NAS use cases -- for SPI flash boot scenarios and system on fast USB storage it's a different story).
msev Posted January 23, 2017 Posted January 23, 2017 Would 10/10 internet be fast enough? What about 10/1? (I need to check what kind of speeds i have). So this USB3 GbE dongle creates a USB3 port and intefaces with gigabit port on SBC? So what is the point of this addon? Will we be able to reach the speed of Odroid XU4 with the use of H5 boards?
tkaiser Posted January 23, 2017 Posted January 23, 2017 Hmm... wonders do not happen, H3 and H5 are not USB3 capable and boards with only Fast Ethernet and crappy WiFi aren't that suitable as NAS. In other words: expect performance to be low as hell with actual OPi Zero since network is the bottleneck. So you either add GbE (the link above is there for a reason. To get throughput numbers with this setup!) or wait for a newer Zero that might be made by Xunlong (no idea though, I'm not into their plans and no product manager). This NAS expansion board is interesting due to superior USB-to-SATA bridges used there, the combination with actual Zero isn't if it's about the NAS use case. But I think I'm repeating me for no reason.
msev Posted January 23, 2017 Posted January 23, 2017 So Odroid XU4 is still best for that use case, would it make sense to use this addon board with it?
Igor Posted January 23, 2017 Posted January 23, 2017 So Odroid XU4 is still best for that use case, would it make sense to use this addon board with it? For this price tag it make sense. XU4 costs much more, consume more energy, it's loud, ... and also not the only choice left for this job.
tkaiser Posted January 23, 2017 Posted January 23, 2017 So Odroid XU4 is still best for that use case, would it make sense to use this addon board with it? Neither/nor. Why not reading through comments in link from 1st posting? Doing some research on your own? Define the meaning of 'best'? If Xunlong comes up with a new GbE equipped zero for $15 then this in combination with the NAS Expansion board can be great for many use cases. In other situations ARM boards with Marvell SoCs are 'best' and for other NAS situations x86 boxes. It depends and it's necessary to define use cases and get into technical details (this forum is full of BTW!)
msev Posted January 23, 2017 Posted January 23, 2017 I did research and IMO XU4 is the best regarding price performance with OpenMediaVault (because its ARM and uses less power then x86, thats why is better, and tutorials are available, they state read and write speeds of 80mb's). Marvell boards are more expensive (110usd). I don't understand how you circumvent the lack of USB3...in that case read and write speeds will never be as good as XU4. Use case is home NAS for media + databases from homeautomation + security camera stuff.
tkaiser Posted January 23, 2017 Posted January 23, 2017 Use case is home NAS for media + databases from homeautomation + security camera stuff. So even boring OPi Zero could do the job in case your security cameras are able to encode in h.264 or better? It always depends on the use case. If I need a backup NAS for a few Macs for example I can use any A20 board since TM is slow when backing up and real performance is only needed when restoring: http://linux-sunxi.org/Sunxi_devices_as_NAS#Influence_of_the_chosen_OS_image_on_NAS_performance If I only want to watch movies, then most probably Fast Ethernet is enough since most media formats are encoded (for a reason!). If I want to impress others I combine a fast and expensive SSD with a fast device and produce useless benchmark numbers that look nice. If I want to setup redundant storage for a whole company then I choose x86 boxes with a couple of HDDs (then the 'system in question' doesn't matter at all since those boxes need redundant +1000W PSUs anyway), SSD caches and at least 10 GbE Ethernet. Again: use case. Regarding 'Marvell boards' and your price tag: https://www.kickstarter.com/projects/874883570/marvell-espressobin-board (it gets boring since all this information is already available, just click on the link in the first post above).
msev Posted January 23, 2017 Posted January 23, 2017 So for my use cases, streaming 1080p video and stuff like that, would it work smooth without stuttering? Maybe even to multiple TVs at the same time? Even a Orange Pi zero or Orange Pi PC would be good enough? Would i be able to set up RAID? So that data would stay safe if one disk died?
tkaiser Posted January 23, 2017 Posted January 23, 2017 So for my use cases (streaming 1080p video and stuff like that, would it work smooth without stuttering?) even a Orange Pi zero or Orange Pi PC would be good enough? Would i be able to set up RAID? So that data would stay safe if one disk died? '1080p video' is resolution but doesn't tell anthing regarding bitrates (depends on encoding/codec used). So you have to check how high bitrate requirements of your media files are. Please note: most people confuse codecs with file/container formats. If you have files lying around with names ending on .mp4 that still doesn't tell anything: https://en.wikipedia.org/wiki/MPEG-4_Part_14#Data_streams The good news: bitrate requirements for most media formats are rather low so even Fast Ethernet should suffice (since this stuff is meant for streaming and the usual encoders are configured to not exceed a specific bitrate anyway -- in case 'the scene' needs higher bitrates than allowed the encoder partially decreases image quality to remain below the bitrate treshold). Fast Ethernet means 8 MB/s at least. With current cpufreq settings in Armbian images actual values might be lower since cpufreq might remain at the lower 240 MHz most of the time. Switching to performance governor might help: http://linux-sunxi.org/Cpufreq#The_.22performance.22_governor But even the 5 MB/s reported by @jimandroidpc should be sufficient for most media files (please remember that there are many Android TV boxes with only Fast Ethernet around and the bitrates that work reliably over crappy 2.4GHz Wi-Fi have to be even lower). I won't comment on RAID used with SBC since I did it too often already. TL;DR: RAID has nothing to do with 'data safety', this is only about availability (business continuity) and people constantly fool themselves with RAID (if you're interested in my opinions a web search for 'tkaiser crap raid availability' or something like that should help). Again: I really hope that Xunlong already prepared a new board with Gigabit Ethernet and Zero form factor since with current Zero the NAS Expansion board seems somewhat strange (the JSM578 is capable of SAT and therefore also SMART, supports TRIM/UNMAP and also UAS which is nice since Allwinner SoCs can make use of UAS even with USB 2.0 when running mainline kernel. Normally you find this feature only on USB3 capable devices). But JMS578's potential is wasted on a board with only Fast Ethernet and crappy Wi-Fi. 2
technik007_cz Posted January 23, 2017 Posted January 23, 2017 This nas expansion is great idea when ssd drive lifetime and performance is much better than sd card. If it will be released with proper plastic case I will buy it. (performance bottleneck of usb sata converter I am leaving without comment. This is cheap device area and I do not expect this item flying for this price.) Sent from my Redmi Note 2 using Tapatalk
malvcr Posted January 23, 2017 Posted January 23, 2017 This is my first post here ... Recently I acquired some OPIZ, and I have been working with them for having cyphered storage. As my first attempt was using Raspberry Pi Zero machines, to jump to OPIZ machines was an improvement ( around 20 times the total speed ). Of course that when using the Fast Ethernet we have a clear bottleneck, but it depends on the usage. Also, it is possible to use g_ether (Ethernet over USB and to move more to the USB 2.0 limit that it is clearly higher than the Fast Ethernet one). In my case I have been linking two OPIZ with USB reprogramming the OTG to work as a client (mode 2), so one of them has a web server and the other a database and a cyphered database located in an USB hard disk. For that I am using Wester Digital PiDrives, that are native USB disks. However, could be useful to have more storage options, and I think that the mentioned expansion board is a good solution because it opens possibilities. One obvious advantage is to eliminate many cables that can make your life complicated. Not everything is about a lot of data in very short periods of time. There are plenty of cases where USB 2.0 based storage is more thank enough, and as the OPIZ H2+ inherits the separated channels for all its USB 2.0 ports from the H3 chip, it is really a very powerful contender in its league. Of course than to have USB 3.0 "separated" channels would be wonderful, but I think this moves the line to the next machine level we can work with if we need to do so. Oh, I prefer not to have internal hubs or switches in these small machines (as the Raspberry do) ... to be confident on the total bandwidth. I think this is the main cause of the dramatic performance difference between the Raspberry Pi Zero and the Orange Pi Zero.
tkaiser Posted January 24, 2017 Posted January 24, 2017 performance bottleneck of usb sata converter I am leaving without comment. This is cheap device area and I do not expect this item flying for this price. Well, but Xunlong put on this $7 Expansion board two of the best USB-to-SATA bridges currently available: JMS578. Performance with USB2 devices should be comparable to JMS567 (see some results here -- please note the newly submitted results with A64 SoC since I would suspect the same would be possible with H5) and with USB3 hosts these beasts really rock. But I doubt we'll ever see an USB3 capable Zero variant and I also doubt that USB3 Superspeed works over a GPIO header (but I'm an electronics noob so might be wrong here) Oh, I prefer not to have internal hubs or switches in these small machines (as the Raspberry do) ... to be confident on the total bandwidth. I think this is the main cause of the dramatic performance difference between the Raspberry Pi Zero and the Orange Pi Zero. Well, RPi Zero has no internal USB hub and a rather inefficient ARMv6 CPU core. But it's true: with every Raspberry around both I/O and network bandwidth are limited due to network hanging off the USB bus and the SoC only capable of one single USB2 connection to the outside. H2+/H3/H5 provide 3 real USB2 host ports and 1 OTG port that can be used in host mode and then performs just a little bit lower. I'm a bit curious regarding the PiDrive since when I did some research on it over a year ago it doesn't looked that promising. Could you run an iozone benchmark after switching the OPi Zero to performance governor with the 3 calls from here: http://linux-sunxi.org/USB/UAS#Testing_3_different_external_enclosures And also query drive health using 'sudo smartctl -a /dev/sda' (requires an 'sudo apt install smartmontools' first of course).
msev Posted January 24, 2017 Posted January 24, 2017 So what would be then the preffered way of keeping data safe on a "Opi NAS" with 2 disk for example. Just copying manually critical data on both disks and then if one disk dies I still have the other one?
tkaiser Posted January 24, 2017 Posted January 24, 2017 So what would be then the preffered way of keeping data safe on a "Opi NAS" with 2 disk for example. Just copying manually critical data on both disks and then if one disk dies I still have the other one? If your house burns down there is no other one. The only way to achieve 'data safety' is to become familiar with the concepts needed and then implement them based on the level of 'safety' needed. Modern filesystems allow the creation of snapshots and sending them over to other disks or remote locations (then your house can burn down and at least your data survives that). With old and boring filesystem tools like rsync could be used. But without wasting some time on basics/concepts you can neither implement anything nor gain any level of 'safety'. That's why RAID is so popular: easy to implement, doesn't protect from most reasons of data loss/corruption, good feeling ('at least I've done something and wasted a whole disk for this feeling').
msev Posted January 24, 2017 Posted January 24, 2017 I don't understand why you are against RAID (haven't googled yet your opinon), but RAID 1 disk mirroring should save me if 1 disk dies...I don't get it..still better to have photos on laptop, and on raid 1 two disks on OPI Nas. And why would snapshots be better?
tkaiser Posted January 24, 2017 Posted January 24, 2017 haven't googled yet your opinon Great, so let's start repeating http://www.justfuckinggoogleit.com stuff over and over again to get this forum flooded with useless posts? Data mirroring (RAID-1) without using recent filesystems that ensure data integrity is plain stupid. Reasons can be found on the net and on this forum eg. https://forum.armbian.com/index.php/topic/2921-software-raid-on-opi-pc/ Let's stop here. Everything is already said thousand times. Let's stop writing and start reading (do some Google-Fu before). And please try out the following to get the idea how a data mirror correlates with 'data safety'. Just do a 'sudo rm -rf /' on your SBC and then educate yourself what snapshots are and how they prevent from a situation where everything is lost when using stupid RAID modes. 1
msev Posted January 24, 2017 Posted January 24, 2017 In the thread that you linked to the guy wanted to do Raid 1 on the flash storage (eMMC and SD card), while I would do it on a HDD, so I don't know how relevant that thread is to this situation? Btw would it make any sense to use this Opi NAS hat with Odroid XU4?
bzfrp Posted January 24, 2017 Posted January 24, 2017 Fast Ethernet means 8 MB/s at least. With current cpufreq settings in Armbian images actual values might be lower since cpufreq might remain at the lower 240 MHz most of the time. Switching to performance governor might help: http://linux-sunxi.org/Cpufreq#The_.22performance.22_governor Sorry for my stupid question. Instead of switching to performance governor to fix the frequency, can we disable cpufreq in the kernel and set the desired frequency in uboot? Is there a difference between these two methods? Thanks.
malvcr Posted January 24, 2017 Posted January 24, 2017 (edited) Hi tkaiser To run these tests takes some time :-) I was reading the previous post about the PiDrive ... 10 MB/s it is not real. I have two sets of numbers here taken from a fresh PiDrive on a recently installed Orange Pi Zero and an SSD (Kernel 3 series, so I can't check the UASP right now ... although I will do it these days because it is important). According with my previous tests, these numbers (10 MB/s) are more for a Raspberry Pi with its inherent bottleneck, not for an Orange Pi, or because some network was in between. My guess is that the machine is the problem, not the disk neither its integrated circuits. About the second set of tests ... they are interesting because they belong to a SanDisk Z400s SSD with LUKS on an INITIO INIC 1618 based USB 2.0 Seagate bridge (I need to retest this without LUKS, but it seems that the bridge I am using for that it is "terrible" ... I don't think the SSD is the responsible for the slow numbers). And, in this case, the integrated Orange Pi card looks to be a wonderful idea. Also, while making the tests I was trying to replace the USB-SATA bridge with another one on the Z400s, but it reported a different geometry to Linux, making the LUKS to fail. In this case, to have USB hard disks it is nice, because the hardware combination it is married and that problem won't exist. For me both, the Xunlong and the WD solutions are good ones, maybe for different types of solutions (not all problems are nails neither the tools to resolve them are hammers). Z400s Data: iozone -a -g 4000m -s 4000m -i 0 -i 1 -r 4K random random kB reclen write rewrite read reread read write 4096000 4 28300 28715 19252 19158 iozone -a -g 4000m -s 4000m -i 0 -i 1 -r 1024K random random kB reclen write rewrite read reread read write 4096000 1024 28202 29010 18720 18589 iozone -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 2000m random random kB reclen write rewrite read reread read write 2048000 4 6954 0 4856 0 662 1906 PiDrive (314) Data: iozone -a -g 4000m -s 4000m -i 0 -i 1 -r 4K random random kB reclen write rewrite read reread read write 4096000 4 34354 34557 32271 32250 iozone -a -g 4000m -s 4000m -i 0 -i 1 -r 1024K random random kB reclen write rewrite read reread read write 4096000 1024 34407 34329 31539 31500 iozone -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 2000m random random kB reclen write rewrite read reread read write 2048000 4 8451 0 8049 0 114 992 Smartctl: smartctl -a /dev/sda smartctl 6.4 2014-10-07 r4002 [armv7l-linux-3.4.113-sun8i] (local build) Copyright © 2002-14, Bruce Allen, Christian Franke, www.smartmontools.org /dev/sda: Unknown USB bridge [0x1058:0x259c (0x1005)] Please specify device type with the -d option. Use smartctl -h to get a usage summary Using "sat" parameter: smartctl -a -d sat /dev/sda smartctl 6.4 2014-10-07 r4002 [armv7l-linux-3.4.113-sun8i] (local build) Copyright © 2002-14, Bruce Allen, Christian Franke, www.smartmontools.org === START OF INFORMATION SECTION === Device Model: WDC WD3140LMCW-11D9GS3 Serial Number: WD-XXXXXXXXXXXX (replaced on purpose) LU WWN Device Id: 5 0014ee 65bf0cc64 Firmware Version: 01.01A01 User Capacity: 314,071,769,088 bytes [314 GB] Sector Sizes: 512 bytes logical, 4096 bytes physical Rotation Rate: 5400 rpm Form Factor: 2.5 inches Device is: Not in smartctl database [for details use: -P showall] ATA Version is: ACS-3 T13/2161-D revision 5 SATA Version is: SATA 3.1, 1.5 Gb/s (current: 1.5 Gb/s) Local Time is: Tue Jan 24 17:58:40 2017 CET SMART support is: Available - device has SMART capability. SMART support is: Enabled === START OF READ SMART DATA SECTION === SMART Status not supported: Incomplete response, ATA output registers missing SMART overall-health self-assessment test result: PASSED Warning: This result is based on an Attribute check. General SMART Values: Offline data collection status: (0x00) Offline data collection activity was never started. Auto Offline Data Collection: Disabled. Self-test execution status: ( 0) The previous self-test routine completed without error or no self-test has ever been run. Total time to complete Offline data collection: ( 6720) seconds. Offline data collection capabilities: (0x7b) SMART execute Offline immediate. Auto Offline data collection on/off support. Suspend Offline collection upon new command. Offline surface scan supported. Self-test supported. Conveyance Self-test supported. Selective Self-test supported. SMART capabilities: (0x0003) Saves SMART data before entering power-saving mode. Supports SMART auto save timer. Error logging capability: (0x01) Error logging supported. General Purpose Logging supported. Short self-test routine recommended polling time: ( 2) minutes. Extended self-test routine recommended polling time: ( 79) minutes. Conveyance self-test routine recommended polling time: ( 5) minutes. SCT capabilities: (0x3035) SCT Status supported. SCT Feature Control supported. SCT Data Table supported. SMART Attributes Data Structure revision number: 16 Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail Always - 0 3 Spin_Up_Time 0x0027 130 124 021 Pre-fail Always - 2500 4 Start_Stop_Count 0x0032 100 100 000 Old_age Always - 86 5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail Always - 0 7 Seek_Error_Rate 0x002e 200 200 000 Old_age Always - 0 9 Power_On_Hours 0x0032 100 100 000 Old_age Always - 270 10 Spin_Retry_Count 0x0032 100 253 000 Old_age Always - 0 11 Calibration_Retry_Count 0x0032 100 253 000 Old_age Always - 0 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 22 192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age Always - 5 193 Load_Cycle_Count 0x0032 199 199 000 Old_age Always - 3903 194 Temperature_Celsius 0x0022 104 100 000 Old_age Always - 39 196 Reallocated_Event_Count 0x0032 200 200 000 Old_age Always - 0 197 Current_Pending_Sector 0x0032 200 200 000 Old_age Always - 0 198 Offline_Uncorrectable 0x0030 100 253 000 Old_age Offline - 0 199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age Always - 0 200 Multi_Zone_Error_Rate 0x0008 100 253 000 Old_age Offline - 0 SMART Error Log Version: 1 No Errors Logged SMART Self-test log structure revision number 1 No self-tests have been logged. [To run self-tests, use: smartctl -t] SMART Selective self-test log data structure revision number 1 SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS 1 0 0 Not_testing 2 0 0 Not_testing 3 0 0 Not_testing 4 0 0 Not_testing 5 0 0 Not_testing Selective self-test flags (0x0): After scanning selected spans, do NOT read-scan remainder of disk. If Selective self-test is pending on power-up, resume after 0 minute delay. As a reference, this is the Z400s data (some because it is very long): === START OF INFORMATION SECTION === Device Model: SanDisk SD8SBAT128G1122 Serial Number: XXXXXXXXXXXX (replaced on purpose) LU WWN Device Id: 5 001b44 4a44dd3c2 Firmware Version: Z2320000 User Capacity: 128,035,676,160 bytes [128 GB] Sector Size: 512 bytes logical/physical Rotation Rate: Solid State Device Form Factor: 2.5 inches Device is: Not in smartctl database [for details use: -P showall] ATA Version is: ACS-2 T13/2015-D revision 3 SATA Version is: SATA 3.2, 6.0 Gb/s (current: 1.5 Gb/s) Local Time is: Tue Jan 24 17:14:27 2017 CET SMART support is: Available - device has SMART capability. SMART support is: Enabled (...) SMART Attributes Data Structure revision number: 1 Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 5 Reallocated_Sector_Ct 0x0032 100 100 000 Old_age Always - 0 9 Power_On_Hours 0x0032 038 100 000 Old_age Always - 38 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 450 166 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 1 167 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 0 168 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 2 169 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 228 170 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 0 171 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 0 172 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 0 173 Unknown_Attribute 0x0032 100 100 --- Old_age Always - 1 174 Unknown_Attribute 0x0032 100 100 000 Old_age Always - 332 187 Reported_Uncorrect 0x0032 100 100 000 Old_age Always - 0 194 Temperature_Celsius 0x0022 068 100 000 Old_age Always - 32 (Min/Max 0/45) 199 UDMA_CRC_Error_Count 0x0032 100 100 000 Old_age Always - 2 230 Unknown_SSD_Attribute 0x0032 100 100 000 Old_age Always - 5 232 Available_Reservd_Space 0x0033 100 100 004 Pre-fail Always - 100 233 Media_Wearout_Indicator 0x0032 100 100 000 Old_age Always - 80 241 Total_LBAs_Written 0x0030 253 253 000 Old_age Offline - 82 242 Total_LBAs_Read 0x0030 253 253 000 Old_age Offline - 46 (...) About the RAID discussion. I could use RAID (whatever number) inside a critical expensive centralized system, but for so small machines, I think it is much better to use DRDB. I can have two completely independent computing systems making automatic physical level backup between them ... IF ... the quantity of data it is reasonable, because they would need to work with USB 2.0. To put so many redundant systems INSIDE so small machines doesn't look to be a good idea; anyway, the external disk must be USB, so you can chain the machines using also USB and hence the difference in speed doesn't need to be so abysmal. And using these NAS expansion boards seems interesting, because you can have two semi-integrated storage nodes, one being a backup from the other one... but again, for data moving at USB 2.0 speed needs (small files, sensor data, not so heavy used databases ... ). Edited January 25, 2017 by tkaiser Formatting fixes, added spoiler tags
tkaiser Posted January 25, 2017 Posted January 25, 2017 Sorry for my stupid question. Instead of switching to performance governor to fix the frequency, can we disable cpufreq in the kernel and set the desired frequency in uboot? Is there a difference between these two methods? Let me explain the problem first: With Fast Ethernet or Wi-Fi the CPU cores run into an chicken-egg problem: Our cpufreq settings let H2+ remain at the lower clockspeeds until something important has to be done. Transferring data through the network is bottlenecked by either Fast Ethernet or the crappy/cheap Wi-Fi so there's not that much to do which leads to the SoC only raising cpufreq from time to time instead of running all the time at the upper limit (1.2 GHz). On the other hand with these cheap SoCs both network and storage performance is dependent on CPU clockspeed. So without adjusting cpufreq to the upper limit NAS transfers will be thwarted and by using performance governor (or any other method to configure CPU cores to run at 1200 MHz) throughput numbers are lower than necessary. Please educate yourself about different options and interaction between u-boot, kernel and userspace settings: https://forum.armbian.com/index.php/topic/1728-rfc-default-settings-for-nanopi-neoair/ Recommended way is to adjust cpufreq-utils settings (or a script adjusting the settings 'on demand' or even trying out whether ondemand governor with appropriate settings (important -- without the IO related switches it will never work) gives better numbers: http://linux-sunxi.org/Cpufreq#The_.22ondemand.22_governor (information partially outdated -- I didn't know better back at that time) I was reading the previous post about the PiDrive ... 10 MB/s it is not real. Thanks for the numbers. It looks really like the USB bridge between the SSD and your board is horribly slow. Regarding UAS I don't believe the PiDrives are capable of doing so (would be interesting when you do the test with mainline kernel but since WD chose to use an own vendor/product ID even if the bridge used is UAS capable kernel won't recognise it). Another concern with WD drives is high load cycle count (just do a google search for 'LCC WD') but this looks ok-ish in your SMART output (14 times per hour parking/unramping the heads might be a reasonable result based on usage pattern).
tkaiser Posted January 25, 2017 Posted January 25, 2017 Almost forgot: On all the small IoT boards like OPi Zero (that are happily mis-used) we use pretty low DRAM clockspeeds of just 408 MHz. This also slightly affects 'NAS throughput' but increasing DRAM clockspeed won't help that much while increasing consumption a lot (and board temperature for reasons I still don't understand). BTW: for reasonable NAS throughput/latency tests I strongly recommend the use of Helios Lantest with appropriate settings (details and correlation with real-world workloads explained here: http://www.helios.de/web/EN/support/TI/157.html -- people who don't want to get into details of course don't need to test anything)
Magnets Posted January 25, 2017 Posted January 25, 2017 This board looks useful but as people have already said, they should make some low end boards with gigabit ethernet. What's the point if you only have 100mbit. Does it come with the standoffs to mount it? Not sure why someone else in the thread was having problems with 100mbit, I can easily saturate it on an Opi PC writing to usb.
Recommended Posts