Manufacturers continue to play with NAS configurations in their fight for marketshare. This one combines media serving, gigabit Ethernet, wireless and BYOD features. What do you think?

Although I think you guys did a fair review of, at the time, a mediocre product, I'm utterly stumped as to why the review was performed on hardware RevA unit!

RevB has been selling now for at least 4months and is the only version I've seen on retailers shelves. The RevB box improves on many of the RevA's shortcomings. Not the least of which is a faster Freescale PowerPC based CPU. On the feature side, the RevB supports FTP access, user creation, and emailing alerts/logging... all features that were noted in the review as being hidden and possibly broken. In addition it DOES support NTFS mounted partitions, though in read-only mode. The RevB is also imminently more hackable via a wget hack than what Jim had to go through with the RevA.

[disclaimer] I don't mean to come across as a DSM evangelist, but I do own one (RevB). It's not perfect, but in my situation, and for the price, it worked for me. [/disclaimer]

I just have to ask... How much value is this review going to be for someone looking to purchase a NAS if it's based on an older hardware revision, with less features that's scarcely available in channels anymore, if at all?

Far be it from me to dictate anything... But may I make a suggestion? Tim, and Tom's Networking by extension, has a long history of making updates to reviews when new firmware or updates become available. Is it possible to put a RevB DSM-G600 though it's paces and update this review?

I know, personally, I'd like to see how a RevB DSM-G600 stacks up to the competition; but more importantly I think it would give a prospective NAS buyer a more current, fair comparison on which they can base their future decisions.

I would just like to agree with QBiN. The version "B" running firmware 1.01 is a vast improvement over the A version.

Please note that the "B" version also has FTP capabilities, and if I recall correctly (the box is not in front of me right now), it has the capability to do a scheduled automated FTP for backup(not a feature that I use, though).

And as to your analysis of the cost of the unit, a lot of people have spare hard drives lying around that, for all practical purposes, are "free".

Another plus of this, as compared to boxes like the Netgear SC101, is that it supports SMB directly, so it will serve files to Windows, Apple, and Linux boxes (a definite plus).

Finally, the comment that this does not offer DHCP or firewall over the wireless network. That is just plain STUPID. This is an "Access point" and NOT a "wirelss router." Show me another box that is just an access point that has that stuff.

Another note on file systems... I had to pull apart a USB enclosure to liberate a drive to use in my DSM-G600. So, I had this empty USB enclosure, and I had a couple of spare CD-Rom drives lying around. I make a USB CD-Rom drive, and plugged it into my DSM-G600. The box mounted the drive, and presented the CD-Rom filesystem as a new share in a matter of seconds. This is pretty cool, and I can think of some interesting uses for being able to remote-mount a CD-Rom. I will have to try a DVD drive someday soon.

Here is my list of the problems with the unit:

1) Performance is a little on the slow side, but typical for this class of device. I believe that there is some overclocking potential in the processor. I will play with this one day.
2) Does not spin down hard drive (but this may be fixed later).
3) Fan is loud. I may add a thermal circuit to control this one day too.
4) I have found a "microsoft office" bug. Copy a word or excel file to the unit. Open it, edit it, and click on the "save" button. Your computer will lock up for a minute or two. I have seen this using Office 2003 on XP and Office 97 on 98SE.

Good things about the unit:
1) If you have a hard drive lying around, it is only $150.
2) Works with all operating systems, including Linux and Mac
3) Manufacturer seems intent on upgrading the firmware.
4) Potential for some cool hacks.
5) Can mount and share a USB CD-Rom
6) Easy to set up.

You mention the Rev. B uses a PowerPC processor, could you enlighten me as to which one? Thanks.

Hi Guys, thanks for the feedback.

As far as the Rev A/B issue, I can only test what D-Link sends. We requested a review unit, and reviewed what we were sent.

Maybe I didn't get the point across on the wireless capabilities, but what I was trying to say is that you'd get more bang-for-the-buck with most any of the inexpensive access point/routers that are on the market today. As long as you are going to add wireless to your net, you might as well go with something that will give you more features and more flexibility.

Jim Buzbee

I believe it's the Freescale MPC8241 (PowerPC 603e derivative) or a close variant there of. Although, that's from memory last time I installed a drive in mine.

Jim's review of the Iomega StoreCenter had a link to Freescale's page for this CPU:
http://www.freescale.com/webapp/sp [...] H3bTdG8655

Jim, to your point about the eval unit sent to you by DLink. It comes as a surprise to me, if not most of us, that DLink would send an old model. That definitely does not speak well for them and is an obvious blunder on their part.

I have to tread carefully here, because I'm a huge fan of Tim and Jim's work (and have, in fact, emailed Jim to that effect in the past), and I mean no disrespect nor am I intending to call anyone out. That said, given the amount of research and testing that goes into a review, I'm just surprised that the RevA/B issue and it's impact on the review was not even mentioned, if it was known. It still represents an oversight.

Is this the result of different sourcing methodologies (requesting an eval vs. going to the retailer to get a true sample of what buyers are getting)?

As is, for whatever reason (not pointing fingers), the results were outdated as soon as they were released. For that reason, and for the benefit of enthusiasts and buyers, I still feel an update to the article is in order IMHO.

Or maybe there wasn't a Rev B when D-Link sent it to Jim. Everything has to be a blunder, mistake or screw-up.

I am sure Jim didn't write the article in 10 minutes after he received the DSM-G600 this morning.

Typical Duh-Link product, stuff that almost works, stuff that fits the marketing definition of "works" (Rendesvous with our setup page!), and just plain broken stuff. Thanks for the warnings!

Check out this thread:
http://forumz.tomshardware.com/net [...] 20837.html
On 2/27, Tim said they'd begin to try to get a DSM-G600. The RevB has been out much longer than that. I know because I've had one running in my closet since long before that. So while it didn't take 10min, it certainly didn't take 10months either. I guess you didn't know about that thread, though.

Tim & Jim, this reader's opinion serves *exactly* as case in point to what I mentioned earlier. Please consider revisiting this.

I have emailed my marketing contact at D-Link asking why RevA product was supplied for review instead of RevB and will let you know what I learn.

As for revisiting this review, that's unlikely. Our reviews take a good deal of time and there are plenty of other products waiting for review. While I wish that we could review every "Rev" version of a product, that's not practical.

[Rant on]
Manufacturers are constantly updating products, usually not changing product model numbers in a way that's visible other than reading the actual product box itself. This puts online buyers at a distinct disadvantage.

This practice is especially unfair to consumers when important changes to features and/or performance are made in a "Rev" model.

If manufacturers don't think the differences in "Rev" products are significant enough to warrant new model numbers (and the associated marketing costs), then that's how we'll play the game, too.
[Rant off]

QBiN's summary of RevA/B differences covers everything except possible performance differences (thanks for providing it, by the way) and will have to do as a review "update".

I agree with you 100%, Tim.

If DLink is that "asleep at the switch", then they ought to bear the bad press. However, it'd be nice if just a mention was made in the article that a newer Rev and firmware exist if the reader so chooses to look into it on their own.

That's a good compromise. I'll make it so.

Thanks, QBiN!

I was wondering what the speeds were in comparsion to a drive connected to a PC via USB 2.0. Are Gigabit NAS drives faster that a USB 2.0 connected drive?

The reason I ask is that I have an HP 3210xi All in one Printer (which can be networked or USB). It has a memory card ready built into it. When I transfer files over the network ; it is dog slow. When I transfer over USB it is much faster.

Generally, USB 2.0 connected drives are going to outperform even gigabit connected NAS since direct-connected drives don't have the additional network protocol overhead.

So I did some homework and read up on IOzone. For the sake of those interested, I followed the guidelines set by Tom's Networking, listed here:
http://www.tomsnetworking.com/2004/03/02/how_we_test/
And produced some updated Performance number for the RevB.

Before I list them, here's how my setup differs... IOzone was run from an AthlonXP 2100+ with 512MB of RAM with a 3Com 3C980C 100Mb NIC across two daisy chained 100Mb switches to the DLink. I don't have a gigabit NIC to test, so these tests will only reflect 100Mb numbers. The DLink has a 300GB Western Digital (WD3000JB) Special Edition drive w/ 8MB of cache.

So the fact that I'm not using a crossover directly to the NAS may result in less that optimal numbers, because these switches also handle ethernet traffic for my other PC's to/from the internet... but you may be surprised at some of the results.

These results are only for the RevB model. I didn't compare them to any other NAS devices out there because I don't have that raw data. However, I made the scales the same as the ones in the review so you can compare side by side with the ones from the original review.

So without further delay:
(Click on graphs to enlarge)
Read Results:

Even disregarding cached performance, the RevB already shows considerable improvement in reads over the RevA.

Write Results:

The caching effect makes this difficult to see, but it appears the write performance is a bit lower than the RevA.

Now, to for the 128MB File Transfer performance:
Read:

Again, we see clearer indication the RevB is faster in reads than the RevA.

Write:

Write is definitely slower.

My observations:
My setup is less optimal than THG's. More of an average user's setup than a direct crossover would be. I should get WORSE results if the RevA and RevB perform similarly with all else held equal. That said, I'm surprised that read performance of the RevB at 100Mb shows an average improvement of over 20%! Conversely, write performance appears to be down about 5% on average. These are interesting results, to say the least.

I'd like to see how the 1000Mb fares, but that'll have to wait.
Thoughts? Comments? Flames?

Anyone interested in a "Hacking the DSM-G600" post?

What file system are you using? Since EXT3 has some primitive journaling, I would expect it to run slower than EXT2 for writes. Reads should be the same speed for either file system.

Good job, QBin.

My experience in writing large files is that I get 5.5MB/s read and 4.5MB/s write. This is using a WD 160GB with 2MB cache, and transferrling gigabyte-sized files.

Once I played around with FTP, and got 6.5MB/s read.

Wow, QBiN, you've been busy with plots and everything! Thanks for running the tests. Looks like you've been quite thorough.

As long as you are using switches (not hubs), you don't need to worry about other LAN traffic affecting results, since any switch today is non-blocking. This means it will support wire-speed (100Mbps in this case) traffic on all ports simultaneously.

We normally run NAS tests with iozone machine and NAS connected to a switch anyway. Jim used a direct connection for the gigabit tests only because he doesn't have a gigabit switch.

Iozone test results are very dependent on the computer used, especially for cached results. So I wouldn't try to draw too many conclusions from peak (cached) result differences.

But the 128MByte test results should be reasonably apples-to-apples, since they are relatively free of caching effects.

Again, good job, QBiN, and thanks for running the test.

Thanks guys.

Yeah, I get 1~2Mbps faster transfer rates via FTP than SMB. Given that the DLink uses Samba as it's SMB implementation, it's probably just a performance different between the Samba daemon and wu-ftpd.

I believe my DLink's drive is formatted EXT3, so that could explain the slower writes. Good call.