D-Link Forums
The Graveyard - Products No Longer Supported => D-Link Storage => DNS-321 => Topic started by: JBMedeiros on November 09, 2009, 01:12:23 PM
-
Hi.
I bought a DNS-321 last week and put 2 x 1TB Hard Drives.
My congratulations for the product, it is really a great storage with a good system.
I choice D-Link because they say that have a Gigabit ethernet (linksys and netgear dont).
I bought a CAT6 Furukawa Crossover cable thinking i will have fastest transfers.
The best speed i have with DNS-321 are 11MB/s with jumbo packet in 9000.
Please telll me what i have to do to have a really Gigabit transfer (at least 30, 40 MB/s transfers).
My FW is 1.03
-
Gigabit connections are auto-MDX and use all 8 wires in the cable. Use a plain patch cable between the two.
As far as 40mbyte/sec transfers, I'd like to know what you're smoking. :D You can expect them in the 20's...
-
I have tryed Crossover, T568B + T568A, T568B + T568B, T568A + T568A.
Jonh I dont smoke but my conversor says that 1Gb are 128MB. 40MB/s are 32% of a "gigabit".
But i know there are many other limitations like hard disk, memory, transfer bitrates, etc...
The really true is that my DNS-321 dont pass a 100Mbit speed.
-
Well, I get mid 20's for transfers from my DNS-321 at best with jumbo frames and for large files. You should be able to do better than you're doing with it.
-
Hi.
I would recommend you to check out the reviews at smallnetbuilder.com. Both the DNS-323 and the DNS-321 are slower than many other NAS boxes. I can also attest that my Linux file server is significantly faster than the DNS-321. This is a documented and well known limitation most likely due to the relatively slow CPU and small memory of the DNS machines.
On the other hand these boxes are significantly cheaper than other NAS boxes and they sip electricity. My DNS-321 uses 6W on idle and 16W when writing the two 1.5TB HDDs (in mirror). My Linux file server gobbles up more than 60W even when idle.
In my (limited) experience the DNS-321 is faster than a 10/100, and fast enough to play back video from it, although I haven't tried large, HD rips. For backing up large data sets you might be better off with a faster solution - but that will cost more and will probably use more electricity. (Most of the other NAS boxes are glorified Linux boxes just like the DNS, only they run faster CPUs and have more memory. And thus will sip more electricity. And cost more.)
For using it as a simple media / file server and / or backup server of personal computers, where the size of the incremental backups is not that large, the DNS-321 is just fine - IMHO. For anything more demanding you might be better off with a more advanced (=expensive) box.
-
I've achieved 16-18 MB/sec with EXT2 formatted drives in Raid 0. That was just one of the tests I was conducting. After I was done I ended up with EXT3 formatted drives in a RAID 1. This results in a top speed of about 15MB with jumbo frames enabled (on both the NAS and computer network card) but nominal speed is 11-13 MB/sec.
It sounds to me that you have nothing configure wrong.
Like mzpx stated, it's the slow electronics in the NAS causing the bottleneck. Also, it may be a GB connection but it's not certified as a GB connection (does not comply to the standard).
This is still a decent NAS, especially for the price. I wish it did have more RAM and power though. Maybe the next one. Atleast I can access the data on these drives if the NAS fails.
-Joe
-
The device has nay the memory or CPU to handle a significant transfer rate. Does not matter to me. It does the job that I procured it form.
TheWitness
-
Yup, always gotta remember how little this thing costs. Damned thing is reliable so far compared to a RAID5 NAS I had continuously break on me. RAID does not matter if the controller/os portion breaks.
-
Raid5, nasty. Raid1+0 is heaven, and 30% faster with many controllers.
-
Raid5, nasty. Raid1+0 is heaven, and 30% faster with many controllers.
Apparently, you forget that RAID-1 only yields half the capacity of the disks. RAID-5 with a decent number of drives (3 minimum) does much better in terms of storage capacity as well as providing redundancy. You're not going to be running RAID-0+1 with two drives anyway. ;)
-
Apparently, you forget that RAID-1 only yields half the capacity of the disks. RAID-5 with a decent number of drives (3 minimum) does much better in terms of storage capacity as well as providing redundancy. You're not going to be running RAID-0+1 with two drives anyway. ;)
Nope, did not forget to mention it, just hate Raid5 due to the degraded I/O performance when you have a disk that is out of the array. With Raid1+0, you get the benefit's of striping (I/O performance) and redundancy without any degradation when you have a bad disk.
Yes, I admit, you don't get that here though. I use Raid1 due to I don't want to loose data (without me hitting the delete key at least ;))
TheWitness
-
Of course, in normal operation, RAID-5 is just fine, since normally you aren't running with a disk out of the array! :D I don't worry about the issue of losing the disk, that's only a fallback position that should be corrected pretty quickly.
-
For me, where I focus on a very high I/O bandwidth application that writes hundreds of GB every 5 minutes at times, I can not afford an array in a degraded status. For most applications though, it's not a problem. However, if your I/O wait stays near 100%, being degraded, would bomb the server. So, we are both correct ;)
TheWitness
-
I doubt you'll be writing 100's of gigabytes in 5 minutes to this unit. :D
-
No doubt about that!