Performance and Stability Testing
All of this said, how does the device perform? As per my usual reviews, I have been using the device to provide internet connectivity on my home network, since while controlled tests are of merit they rarely reflect the realities of how a device like this may be used by its owner. In terms of stability of the WAN connection I am purely testing how capable the router is when it comes to holding a stable session to the ISP, since the physical component of the connection is handled by an external VDSL Modem.
While I haven’t been testing the device for as long as I would normally like, it does seem to be holding its own in terms of keeping its WAN connection solid. I have included the sync data for comparative purposes, since while the WRT 1900ACS doesn’t have a modem, it does show what level of stability the connection ‘should have’
|Start Time||Stop Time||Downloaded Data||Uploaded Data||Disconnect Reason|
|2016-10-13 19:14:42||2016-10-14 12:18:05||11.1GB
|2016-10-13 09:11:52||2016-10-13 19:13:43||3.9GB||994.9MB||User-Request|
|2016-10-12 14:06:40||2016-10-13 09:10:52||30.1GB||1.0GB||User-Request|
|2016-10-12 14:02:17||2016-10-12 14:02:27||2.5KB||1.1KB||User-Request|
Comparing the two, we can see that ‘authentication’ has been a little patchier than sync, though this is expected, since the first couple of days I was doing some initial testing of config options and set up of the hardware, so the connection has been rock solid since I last stopped making changes that require the device to completely reboot on the 14th. This is supported by the few drops seen from first install on the 12th to the start of the current session all being logged as User-Request by my ISP’s RADIUS platform, since this would have recorded the ‘dying gasp’ sent by the WRT. So, at this stage we are happily pushing 16+ days uptime with nary a blip or loss of performance and considering how I tend to push my connection that’s no mean feat.
Next up we have general Wi-Fi testing, as a rule I prefer to test devices like this in the home or office, where I can replicate what actual users will be experiencing, rather than simply giving lab based results. I’ve expanded the test a little from when I reviewed the EA9500 and started to test different locations and environments to help gauge general performance as distance increases, as well as different and increased amounts of building materials are placed between the router and the client, for the most parts the results should be reasonably predictable, but should also help demonstrate how the wireless chipsets in the device cope with varying levels of signal and noise.
Firstly, I have a few screen grabs from OS X’s wireless access list, to show the various signal strengths and noise ratios encountered.
Testing of throughput was done using iPerf, with a PC desktop, connected over a GigE port, running Linux operating as the server and the client being the same MacBook Pro the wireless data above was taken from. The screen shots above should and indeed do give a reasonable expectation of overall performance based on the measure of throughput that OS X itself is suggesting is supported.
To get an initial reading I tested while in the same room as the router, approximately 5 foot away, with nothing between it and the client, beyond this I would expect either significantly worse or reasonably close levels of performance as more factors are introduced such as distance and various building components, like walls, electrical cabling and plumbing.
Straight away we’re off to a decent start, with TCP clocking in around the 500Mb/sec mark and UDP averaging around the 660Mb/sec mark for 5Ghz. Overall, not bad considering I’m testing this in in the home and not a lab environment, so will rarely if ever see the full benefits of Wireless AC. In terms of the 2.4Ghz frequencies, we’re seeing around the 150-180Mbps mark, again, considering the envronment this isn’t a terrible level of performance, though it may not bode well for the 2.4ghz frequencies when testing elsewhere in my home.
As I move from room to room, the 5Ghz performance is reasonably robust, with speeds generally clocking around the 300Mb/sec mark in the furthest points from the router, suggesting the device should be more than capable of providing good levels of performance for most set ups. The two exceptions to this also happen to be the most hostile to wireless networking, the garage and the bathroom – my garage is full of steel shelving and various piping for the plumbing, which would certainly go some lengths to explain the speed dramatically dropping to sub 50Mb/sec at one point. Similarly, the bathroom, being all hard tile is also going to cause signal loss, which will also impact throughput severely.
Unfortunately, as anticipated the 2.4Ghz does seem to suffer with distance and internal building materials between the client and the router, with speeds dropping to single figures in some extreme cases. Granted, this testing methodology is far from scientific and numerous factors can impact performance in significant ways, however it does show the levels of performance in the home, where a router like this may well be seeing a lot of usage.
Of course, this being a WRT device many of the potential buyers wouldn’t be looking to run the stock firmware, but instead would be looking to flash the firmware with something provided by OpenWRT or its derivatives such as Tomato or DDWRT, Linksys has worked closely with the developers of OpenWRT to ensure maximum compatibility between the firmware and the hardware and this is something we will be looking at in more detail in a future article, since it really does open up the true potential of the WRT range of routers.
Linksys WRT 1900ACS£149.99
- Native open Source support out of the box
- Genernally good 5Ghz performance
- Nice looking unit that is stackable with an 8 Port switch
- Default GUI is very user friendly
- Poor 2.4Ghz performance over distance lets this down.