Trying to find the bottleneck

I could use some ideas on locating the bottleneck in our system.

10M pipe to the Interent --> Cisco3600 --> BH20_1 --> Old CMM --> AP’s, BH20_2 --> CMMmicro --> AP’s

Linktest across BH_1 is 7M both ways
Linktest from 5.7SM-P8 to Advantage AP on Old CMM is 5.3/1.5

Best I have seen is 3.7/1.5 on

If I test from a server directly connected to the Router I get 12M/6M

Seems that I should get more like 5M out of an SM.

We are supecting the old CMM is the culprit, but it’s just a switch (right?). AFAIK it has no IP address, no management, etc, no way to upgrade.

We have another CMMmicro, but before I swap it out I want to make sure it’s worth the time.

Thoughts? Comments?

Are you running Advantage Ap’s in hardware mode, and 2X in the sm?

That’s all we get out of software mode 3.7/1.5

In Hardware mode we got 8.2/3.5 at the sm

sm are p8 forced to hw. did not see a 2x option, was that in the cnut tool?

Sorry Jerry,
These were P-9’s but even in single data rate we were 5.2/1.5

I did have an issue once were I set to many ctrl slots.

have you got ground blocks on those backhauls i recall something about them slowing down traffic.

no ground blocks.

tech support seems to think we should get 5.5 down.

I’m going to replace the cmm and see what happens. everything points to it.

HI Jerry,
Did changing the CMM help your bottleneck?

have not done it yet.

clueless wrote:
Sorry Jerry,
These were P-9's but even in single data rate we were 5.2/1.5

I did have an issue once were I set to many ctrl slots.

Good point about the "too many control slots"...this is a "cousin" to setting the max-distance too far...both add overhead that reduce data throuput:
a) control slots take up data slot 'space'
b) setting max-distance too-far out adds 'space' between the upstream and downstream frames, again reducing the throughput

But...that doesn't sound like what is throttling Jerry's Kids....<grin>

Hi Charles,

I have read several of your comments on things that have been posted, These comments on post are nice but suggestions that might correct the problem are even better :lol:

Ok …hope this is what you mean below…I certainly don’t meant to “Pontificate Without Benefit” (or “just talk to hear myself speak” <grin>)…

1) if you look at the max-range of the AP-SM (found on AP) make sure you aren’t going beyond the necessary range of your furthest SM…

Some people want to “add” to the max range (some actually think that this “turbo-boosts” the range…it doesn’t) - but if you do that you actually “subtract throughput” because the AP will determine (and communicate to the SMs) at what point the AP will “end” it’s transmission…and then how much time will be needed to “wait” for the SMs to begin transmitting back…

This means that if your furthest-out SM is only 4 miles but you set max range for 10 miles that you are actually telling the AP to communicate to the SMs to wait “EXTRA LONG” before the SMs begin to transmit back…that means that the total throughput will be less because the “extra max range” value tells the AP to insert more delay (therefore less time/throughput available for customer bits).


2) Control Slots
Most of the time you don’t have to add more control slots than are already used.

Although there are a max number of 16 control slots, each 2 control slots reduce by “1” the number of data slots (less throughput). The reason that you ‘would’ add control slots (and reduce total resouce for data throughput) is that you have ‘many’ SMs fighting for upstream transmissions…this might occur with many SMs wanting to send many small/quick transmissions (e.g.: Voice over IP or lots of quick IM/text messaging folks)…

Page 192 of the Canopy System User Guide
( … on=1&cat=8)
has a “recommended number of Control Slots” based on number of SMs in your network (assuming “normal” traffic <web-surfing, email, etc.).

Control Slots
For 1-10 SMs you would add 0 Control Slots
For 11-50 SMs you would add 1 Control Slot
For 51-150 SMs you would add 2 Control Slots
For 150-200 SMs you would add 3 Control Slots

So, if you add Control Slots that ‘really’ aren’t needed, then (like the “too-far max-range setting” above) you are wasting time that could be used to send data…

I hope this is what you meant…thanks!

Although your link test from your SM connected to your Advantage AP at the Old CMM site looks good, do you think it would be worthwhile to take a laptop to the site and plug directly into the Old CMM and try a speed test? See what the results are, maybe even try some pings from the CMM Ethernet down to your NOC or even to the Internet. You would be testing your BH link which seems to be great if you are getting 7Mbps both ways.

Or, since its an Old CMM w/ no IP addy, why not do a continuous ping (-t flag) from your NOC to an IP addy of an AP at the Old CMM site? That would force your CMM to query its switching fabric and go up the Ethernet to your AP.

Have you also checked the Ethernet stats of the AP’s at this site? We had some bad cable from Best-Tronics awhile back. First symptom was crawling throughput. Checking the Ethernet stats then revealed an issue with the cabling.

Just some suggestions.

:smiley: That’s what I’m talking about

Were you telling me that ‘my’ post was more in-line with what you consider helpful…or the post from msmith that followed my long lecture :lol:

still learning my way around - thanks for the encouraging ways you used to help me sync-up with this group… :smiley:


My experience has been that setting the max range just past the most distant SM yields poorer performance than pushing the max distance out another 50%.

Any ideas why?

I agree. Our longest shot in around 7 miles. If I set the Max Range, link tests drop about 5-7% for all SM’s. The only Max Range setting that allows me to attain 100% bi-directional link tests is either 14 or 15 miles on 5.7.

Actually, if you understand “what” the max range does, then perhaps this will clarify a “different” problem (but I hesitate to “diagnose from afar”…grin).

The max range setting is like a cheerleader at a football game who yells “Go!” while waiting for the crowd to yell-back “Team!”…that is, if the distance between cheerleader and crowd is small/short, then the cheerleader and the crowd can chant back and forth quickly and there is little “wasted time”.


If the distance between the cheerleader and crowd is “large” (let’s say from the field all the way up to the third tier of the balcony), then the cheerleader will yell “GO!”, but WAIT-LONGER before all the crowd can respond back “TEAM!” in some form of unison…which is now more-stretched-out. If the cheerleader slows-down the pace of “GO!”, then the crowd will similiarly slow-down the pace of “TEAM!”.

If you take this silly analogy we can see one thing about the “timing” and an “unintended consequence” that you might be experiencing…
1) Intended conseqequence: The “slowing down” to include the 3rd tier fo the balcony allows “everyone” to be heard within the “TEAM!” response time…that is good
2) Unintended disadvantage: If you set the max-range distance “TOO FAR”, then you “will” get to hear everybody, but you’ve wasted time that you could have used for more “go”-“team” exchanges…that is, the cheerleader shouts “GO!”, the crowd (who is only sitting in the first few rows) shouts-back “TEAM!”, but then the cheerleader waits to hear if someone is shouting from the 3rd-tier of the bleachers…(which in my highly-contrived scenario they are NOT)…this means that you are wasteing time (throughput) by setting the max-range to listen further/longer when you don’t have anyone out there…
3) Unintended POTENTIAL advantage (?)…PERHAPS by setting the max range further you are allowing the radio time to overcome some multipath/reflections…but I can’t imagine what other benefit that setting max-range “beyond” necessary would have…PLEASE note that I am “grasping” at this one…because I’ve never heard of anyone “suggesting” that the max-range should be set further than the max-range of the furthest-out SM…[UNLESS you are planning to install an SM at a further-out location…but the max-range setting change would only take a moment to make, so I’m not convinced that this would be worth it]

1) you have a fixed set of time to transmit and recieve (2.5 ms)
2) in that fixed time you will send 33 data slots (in my example)
3) if you have SMs that are further out, the max range setting will allow you to “hear” all the far-out SMs, but at the cost of throughput (you will lose some number of data slots…)
4) So if you don’t need ‘far out’ SMs, then set the max range to the, uh, um…‘max range’ you need…that will get you the best throughput for that configuration (all other factors held constant).

Page 195 of the CSUG also reminds people that “you MUST set the (max range) parameter on all other APs in teh clsuter exactly the same”…

My fingers are tired…I think I rambled too long…my apologies, but I hope this helped… :smiley:

NOTE TO CLUELESS: is ‘this’ what you were talking about…more-better???

Hi Charles,

Yes in part. :slight_smile:

Do you or have you operated a canopy wireless system?

I have a very small test network and help support a friends network.

I’m sure you will be a great asset to this forum :!: