According to the article:
The basic test strategy was to copy a file ... as many times as possible,...ending each test run when the notebook computer went into hibernation as a result of near exhaustion of the battery, defined in this case as 5% battery charge remaining."
The benefit of power savings is not realized when you turn your radio back on every three minutes, it is realized when your radio is off for fifteen, twenty, thirty minutes. I think you need to go back and revisit your test design.
Eduardo Fisher
Latest wireless headlines from Network World:
Wireless LANs face huge scaling challenges
Sprint brings more partners aboard for WiMAX rollout
Sprint's WiMax service to include local features
|
Does Verizon's Voyager stack up to the iPhone? |
|
|
5 IT skills that won't boost your salary 
[1,407]
Women 4 times more likely than men to cough up personal info [589]
Japan's 10 funniest tech-related commercials [Videos] [407]
Throwing away a promo CD is "unauthorized distribution"? [1,265]
Adults too quick to dismiss educational video games [682]
Attack of the iPhone clones [Slideshow] [578]
10 things IT needs to know about AJAX [1,258]
This Year's 25 Geekiest 25th Anniversaries [Slideshow] [409]
|
|
No, It's Not
First of all, thank you for the note.
While no test is going to be perfect, expecially when wireless is concerned, this test does in fact point out basic issues with PSM. While you are correct that we'll see a larger benefit by not using the radio for a longer period of time, such kind of begs the question - if you don't need the radio, shut it off entirely, and enjoy the power you've conserved.
But what's the point? How does one test not using something, and why would anyone care about an obvious result?
Moreover, I'm not sure that PSM would help much even in the case of nothing to send vs. CAM under the same circumstances. The radio in either case isn't being handling user traffic, so it's overall contribution to power consumption is negligible in both cases.
A good counterexample here, though, might be the use of one of the asynchronous techniques I mentioned. These could yield excellent savings under both light and heavy loads. I hope to be able to test one or more of these in the future.
Thx. Craig.
Not flawed, but maybe not the whole story
Craig,
I agree that the test was not flawed. However, did you test different combinations of beacon interval and/or DTIM settings on the access point? If so, you might see your results improve (or worsen).
One of the main problems with 802.11 PSM is that it is statically set at the AP for all clients. It has been shown that, as you observed, PSM can have little effect, or even cause higher power consumption, depending on traffic patterns, the round-trip time of TCP traversals, and how packets "line up" in time with the beacon/DTIM intervals. Since the client has no ability to vary its individual power save profile settings, it is at the mercy of the AP settings as to what effect PSM has.
You are correct that in a laptop scenario, the WLAN power savings due to PSM are most likely relatively minor compared to the other subsytems' consumption. In an application such as wireless sensors, though, where the WLAN radio is usually the major system power draw, communications can be infrequent, and one has to deal with a very limited power budget, we have seen that properly tuned PSM settings can significantly reduce the average power consumption of the overall solution. Again, however, the same settings that minimize one client's consumption can have the exact opposite effect on another client.
Thanks. -Michael Blossom
You are correct...
I didn't do any testing varying DTIM simply because of (a) the limited scope and timeframe of the test, and (b) the assumption that users would only rarely attempt to modify this and related parameters and (c) the implementation dependencies that would make the results of such modifications difficult to predict. So I couldn't universally say "do this and you'll see an improvement". It would certainly be worth trying a few experiements here, but the results might be specific to the particular combination of equipment being used, along with the parameters of the particular benchmark. This is basically what you note in your last paragraph.
Who was it that said that complex systems behave conterintuitively? We may have a classic example here. 802.11 is arguably one of the most complex communications protocols ever devised. But such is the price we pay for having to deal with all of the considerations and conditions under which it must operate.
Thank you for the note.
Craig.
Infrastructure
I was wondering what 802.11n access-points did not run with 802.3af? I just finished reviewing at least the largest vendors and they all support 802.11af at N rates in one form or another?
Not sure that's true...
If you look carefully, some operate degraded with .3af power. I've not benchmarked many of these so I can't say exactly what they get. I am wrapping up another Tech Note on some recent testing, though, and I'll post something on that in my blog (http://www.nww.com/community/mathias) shortly. You can see my earlier note on this subject here: http://www.networkworld.com/community/node/25449.
Thx. Craig.
Intel power settings kill performance
I wanted to share my experience with a similar wi-fi power saving feature that degraded wireless performance. I work as a Computer Technician in a medium sized PC repair and sales shop. We resell many major name brand laptops to our clients daily. Last year I began noticing some laptops not connecting fully or properly to our wireless access in our shop during testing. The affected laptops were able to access some sites with ease, but timed out on others. After several cases of this, in and out of our office, I discovered the cause was a setting on the wireless adapter. Many new Intel brand wireless adapters have a power management setting is set to “low” or “medium” by default. This setting directly affects the performance of the wireless adapter, even when plugged into AC power. So now, on each new laptop we resell, our techs have to check this setting on the Intel chips and increase the power to “high” before giving to a client. Or they may have wireless connectivity issues from day one. How odd is that? A default setting that actually makes the wireless connectivity problematic, straight from the factory! So far this is related solely to Intel brand adapters and not others, like Broadcom.
Rodney L.
Computer Tech
Morehead City, NC