Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[boxst]

Hello

What transfer speed should I be expecting?  I'm using homeplugs (but 85mb ones) and I'm getting about 10mb/s for uploading and 20mb/s for downloading.

Thanks,

Steve

[xmetal]

I'm getting about 120mb/s on uploads and about 150mb/s on downloads through the DIR-825 gigabit router (using cat5e cables).  Homeplug, unfortunately, is like wireless...which all depends on the topologies of your network. You'll only be able to achieve 85mb/s under ideal condition, which is never the case in real world houses.

[boxst]

Okay, removed the Homeplug and getting 610 mb/s now.

Thanks.

Steve

[fordem]

Okay, removed the Homeplug and getting 610 mb/s now.

Thanks.

Steve

610 mb/s now ?!?

That would be approximately 61MB/sec significantly over D-Link's advertised speed, which is I recall correctly is just over 20MB/sec - even with the addition of jumbo frame I think that would be a little difficult to believe.

Most of us report speeds in the 20~30 MB/sec range.

[jswashburn]

Are we talking about MegaBytes (MB) or MegaBits (Mb) here?

Maybe it's just my drives (RAID1), but I pull a constant 56Mbit/s (7MB/s) when I copy an ISO file from the NAS to my PC.

Being that both my PC and NAS are connected to a gigabit router (D-Link DGL-4300) using CAT5e, I supposed I can't complain too much. I've tried my best to optimize performance with what I've got. 

[Rodent]

I get about 6 - 7 MB per/s from my DNS323 and Windows XP Pro computers connected to a LinkSys Gig Switch using Cat5e cable.

[dicrouthamel]

I get about 10-11MB/s (mega bytes per second) through a WRT54GS router (wired connection - 10/100).

[fordem]

Well - for me if you say mb/s, I think megabit/sec, and if you say MB/s, I think MegaBytes/sec - how ever you call it - that reported 610 mb/s sounds unrealistic.

6~7 Megabyte/sec is what I would expect on a 100 mbps LAN, at a minimum gigabit should give you double that, closer to triple, and gigabit with jumbo frame should quadruple it - quite seriously you can push in excess of 30 Megabytes/sec through this thing.

[dicrouthamel]

I just tested it by transferring 1.5 GB avi file, so I'm such spitting out what the vista transfer window displays.  Perhaps the previous poster meant 600 KB/s.

[fordem]

My early tests were done with a 2GB zip file and using the SNMP counters on my network switch to provide the measurement data - later ones were done with a little utility called NASTester - written (not by me) to specifically create files, transfer them and time the transfer - there is a link to it somewhere in this forum.

It's important to use a readily available tool - to ensure that the measurements are repeatable and comparable - NASTester gives an average speed for the entire file transfer and can also be used to do a number of transfers and then average the results - other tests such as DUMeter (google it) exist, but I have found that this one reports instantaneous peaks rather an average, in fact, I've recorded speeds of 35MB/sec using it.

[boxst]

The second set of statistics I mentioned were messed up because of some cached information.  On average I get about 80MB/s. 

As per below:



Steve

[jmjansen00]

So that's about 10MB/S then, not bad.

[Tank_Killer]

I get 20Mb/s Reads from the DNS (32Mb/s BUrst for approx 1 min)

And 16Mb/s Writes to my DNS (21Mb/s Burst)

This is Gbic to Gbic with 9k jumbo frames on 5+ gig file sizes.

[Lupismaximus]

I get 16 MB/s when transferring a 2GB file from DNS to PC.  They are connected over GB LAN, using a Dlink DGS-2208.  I'm using Vista 32 bit Ultimate.

It is interesting how everyone is testing this.  If you really want to test this, then do the math.  Time your transfer and understand the file size. 

Also, realize that FTP is going to be different then NFS or SMB.  Also, if you go through your router, you also need to make sure that the router is not connected to the Internet.  The processor can only handle so much!  For this reason, I run my LAN always on a separate switch to ensure the fastest throughput amoungst my endpoints and the DNS

[fordem]

Two things ...

Timing the transfer and doing the math is actually a very inaccurate way - it introduces a "human response time" error - even with a stop watch, could you accurately measure the time taken to transfer a 200MB file?

The NASTester utility I mentioned earlier uses this principle, but automates the process providing easily reproducible results - there's another thread in the forum where I provided a link to a page where it can be downloaded (including source code) if you want to take a look at it.

Second - with today's consumer routers, the router CPU does not paly a part in the switching functions, those are handled by a separate ASIC - I'll try it today, but I seriously doubt that there is any significant difference in throughput to be had by using a separate switch.