Devolo DINrail

I recently installed a Devolo Magic 2 LAN DINrail Add-on adapter to my home's Ethernet over Powerline set up.  I had a specific reason for doing this as I shall explain later.  One thing I found very hard to come by is any form of review of this product and even more difficult to find are benchmarks.  Both of these are important when buying something like this: reviews tell you whether people actually like the product and if, in their opinion, it's worth buying;  benchmarks tell you something about the performance and allow you to assess whether you think it's going to make much of a difference in your intended environment.  Hence, it was a bit of a punt on my part as to whether this device would solve the problem I wanted it to solve.  Bottom-line-up-front, I think it broadly has solved the problem but the performance of my network is perhaps not where one would hope it would be.

Background on Devolo Magic 2 Devices 

I've been a long time user of the Devolo Magic 2 WiFi devices.  These provide a mesh-based WiFi network throughout my home.  I chose to go with a PowerLine-based solution because I live in an older property with thick, solid walls through which WiFi repeater solutions would struggle.  I went with the Devolo products as an upgrade to the previous and very popular TP-Link solution I was using, mostly because the TP-Link devices didn't mesh but also the Devolo devices use the latest G.hn standard.

I found, over time, that the connection in one part of my house was getting so slow that it was causing problems to the devices using that access point.  This could be explained by several reasons but I think the most likely is simply the changes in electrical interference as we added and moved electrical devices around the house.

To understand why I opted for the DINrail solution and the impact this had on my network, you need to understand the layout of my PowerLine solution.

My PowerLine Layout

This is probably best understood through a diagram of my house where I show the 2 floors of the house and rough room layout in green, as follows:

 

The diagram is only subtly different between the first "Before" picture and the second "After" picture.  It represents, very roughly the layout of my house and network: blue colouring for the Devolo devices and Powerline connections; orange colouring for the router device and wired Ethernet connection.  The thin rectangular blue box near the centre of the picture is the consumer unit.

In the "Before" picture, you can see that the Devolo device on the left is the route through which all other Powerline devices reach the Internet, since it's connected directly to my router.  You will also note that three of the Devolo devices are connected to the same ring main and therefore share a trip switch inside the consumer unit.  The fourth Devolo device is connected to a different ring main.  This results in the three, more closely connected, Devolo devices being able to communicate more quickly with each other than they can with the fourth Devolo device.  This comes simply as a result of the fourth device being on a different ring main and the need to hop between them on the DINrail inside the consumer unit.

The "after" picture shows the change in layout after introducing the Devolo DINrail device that is shown as a blue square inside the consumer unit.  Here, the Ethernet connection that previously existed between the main Devolo device (at the left of the picture) and the router has been severed in favour of a new connection between the Devolo DINrail (inside the consumer unit) and the router.

My Goal 

The intention here is to more closely link every ring main of the house with the Internet connection and remove the hop across the DINrail between the Devolo device on the right of the picture and the other devices in my network.  This is one of the strong use cases for installing the DINrail solution since it's able to inject the G.hn signal directly onto each ring main, centrally, and removes the need for each device to broadcast across the whole power network in the house.  Hence, the overall goal here is to substantially increase the performance of the Devolo device on the right of the picture.

Expected Results 

With the above layout in mind and a little bit of knowledge of how things should work in practice.  What I'm expecting to happen, in terms of performance, between the "Before" picture and the "After" picture is that:

1. The Devolo device on the left of the picture would have a slower connection to the Internet.  This results from the fact that it's no longer Gigabit Ethernet connected directly to the router and thus all of its outbound traffic must now flow over the Powerline network.
2. The two Devolo devices in the middle of the house (one upstairs and one in the loft) would be very slightly faster.  This results from the fact that even though their Powerline connection is broadcasting on the same ring main as before, and there is arguably now a little more traffic (from the device on the left of the picture) going over the Powerline connection, that the DINrail should effectively short out the broadcast from the ring main and send it directly to the router.  This, of course, includes the working assumption that these two devices can reach the DINrail first or more easily than they can reach the device on the left of the picture that was formerly the main device linked to the router.
3. The device on the right of the picture that's on a different ring main should see a fairly substantial gain in performance.  This results from the fact that it no longer needs such a high-power broadcast to reach across ring main connections and go completely from one end of the house to the other.  It simply needs to reach back to the consumer unit and its traffic will be directly forwarded over a Gigabit Ethernet connection to the router.

Benchmark Results

I've performed some thorough but fairly rudimentary benchmarks to assess the performance of the network at each access point in the house.  The Devolo Cockpit tool has been used to inspect the connection speed between each Devolo device in the house.  The Ookla speed test tool has been used to assess the Internet connection performance from each of the WiFi enabled Devolo devices in the house.  All benchmarks have been run multiple times, at different times of the day and using a variety of different devices.  The intention here is to minimise any statistical outliers through repeated testing.

	Location	Download (Mbps)	Upload (Mbps)	Idle Latency (ms)	Download Latency (ms)	Upload Latency (ms)
Reference	Wired	456	70	5	70	6
Before	Left	364	69	8	24	32
After	Left	55	52	20	661	1562
Before	Middle	172	67	9	366	31
After	Middle	107	68	16	301	396
Before	Right	26	11	10	740	5785
After	Right	90	71	16	458	1035

Guidelines for interpreting the table:

• All figures measure an average performance over time 
• The reference figure measures a wired connection directly into my router
• Before figures measure performance prior to installing the DINrail
• After figures measure performance after installing the DINrail
• Left refers to the WiFi Devolo device on the left of the picture 
• Middle refers to the WiFi Devolo device in the middle of the picture 
• Right refers to the WiFI Devolo device on the right of the picture
• Coloured blocks help identify which results can be directly compared to each other
• Download and upload speeds are measured in Mbps where higher is better
• Latency figures are measured in ms where lower is better 

Conclusions 

Earlier, in the Expected Results section, I gave a 3-point overview of what I'd have expected to happen as a result of installing the Devolo DINrail into my home's Powerline set up.  The actual results were a little different to that which I had expected.  Unfortunately, the results are worse although overall the reason for installing the DINrail adapter (to increase the performance of the Devolo device on the right of the picture) was achieved.  Revisiting my 3-point overview of expected results, here's what actually happened:

1.  The Devolo device on the left of the picture had a significant drop in performance.  This is, perhaps worse than I was expecting with an average download speed around 15% of the previous set up, a similar upload speed and considerably and somewhere over 1.5 orders of magnitude worse latency.
2. The Devolo device in the middle of the house is slower.  Its average download speed has dropped to around 66% of the previous setup with a similar upload speed and similar latency.  This is completely contrary to my expectation where I thought performance may have a slight increase.
3. The device on the right of the picture has seen a considerable increase in performance with download speeds up 350%, upload speeds up 650% and a substantial reduction in latency.  This is pretty much exactly where I was expecting to land for this device.

The overall conclusion here must therefore be that the Devolo DINrail installation has achieved the goal that I set out.  It has:

• diminished the performance of the 2 devices in the middle of the house very slightly but not to the extent that anybody reasonably notices in day-to-day usage
• very highly diminished the performance of the device at the left of the house which is now only suitable for relatively light usage; fortunately we are able to cable connect devices directly to the router here which alleviates this problem somewhat
• substantially increased the performance of the device at the right of the house, turning it from unreliable, slow and relatively unusable to something that works very well

In addition to this, the overall summary is therefore that the DINrail hasn't necessarily improved performance of my network.  What it has done is balanced the performance more evenly across the network.  Thus, a sacrifice of performance has been made in some areas in order to enhance the performance in another.