CCNP Wireless Design (300-360)

The following information is for anyone that is looking at or is currently studying for the CCNP Wireless Design (300-360) exam.

Information about the CCNP Wireless exams can be found here  as well there is a CCNP Wireless study group

The following material below is what I would recommend for anyone looking at studying for this exam

Study material:

CWNP: Certified Wireless Design Professional study guide (CWDP-302)  https://www.cwnp.com/certifications/cwdp

Cisco CCNP Wireless CUWSS Quick guides

  • Old book, yes but so is Cisco wireless design methodologies

Cisco E- learning

Cisco deployment guides based on the code version of the exam

In Nov 2018, I sat the CCNP Wireless Design (300-360) exam and passed it, I finished the exam with plenty of spare time. The exam had typical Cisco question, where it could have two right answer but the correct one depends on what document you read.

The exam was based on a lot of Cisco best practices and wireless standards, if you have been doing Cisco Wireless or wireless design in general for some time you may find the exam not that difficult. If you haven’t I suggest reading the material mentioned above and visiting the CCNP Study group linked above.

Advertisements

How many clients can we have per radio?

A very common question we wireless professionals get asked is. How many clients per AP can we have on this AP?

To help us determine and better answers this question there are some excellent resource available that I have linked below. Highly recommend becoming familiar with them.

Capacity Planner by Andrew Von Nagy at Revolutionwifi

Forecasting AP Capacity by by Andrew Von Nagy at Revolutionwifi

Wi-Fi Throughput by Devin Akin at Divergent Dynamics

Ekahau Site Survey Pro (any of the latest version) need to pay for software

Client information by Mike Albano at mikealbano.com

To determine client capacity, we need to understand what is actually possible, as we have some major limiting factors with our wireless networks “Air-time” and the half duplex nature of the medium.

Capacity calculators like the ones above are used to determine how many AP’s/Radio’s we need based on the amount/types of clients and their throughput requirements.

To get a general understand of how many clients a radio can support I will show some examples using Devin Akins efficiency rules (linked above) “MCS rate x ~ 40% / # of users = per device throughput.

Following examples are based on zero 802.11 and non-802.11 source of interference (perfect world).

Example 1:

802.11n 2×2:2 AP with 20MHz channel with 50/30/10 802.11n 1×1:1 clients at an RSSI -64 that gives them a MCS7 data rate of 72Mbps with a short guard interval at medium/large use.

Data Rate= 72Mbps

40% efficiency= medium/large use

Total throughput capacity= 29Mbps

/50 1×1:1 client devices =570Kbps/client

/30 1×1:1 client devices =650Kbps/client

/10 1×1:1 client devices =2.9Mbps/client

Example 2:

802.11n 2×2:2 AP with 20MHz channel with 50/30/10 802.11n 2×2:2 clients at an RSSI -64 that gives them a MCS7 data rate of 144Mbps with a short guard interval at medium/large use.

Data Rate= 144Mbps

40% efficiency

Total throughput capacity= 58Mbps

/50 2×2:2 client devices =1.152Mbps/client

/30 2×2:2 client devices =1.92Mbps/client

/10 2×2:2 client devices =5.76Mbps/client

The following examples show the affect that device capability & number has to overall throughput per client.

Wireless networks 99% of the time contain different chipsets, understanding your device types is a critical component to your wireless network. Mike Albano site has an extensive list of different client chip sets, if the device is not listed, it mentions how to capture the information.

A very common way marketing spin their AP’s is with descriptions like: “Our 802.11n 2×2:2 APs can support 500 clients with a data rate of 300Mbps”. Which is just not possible.

Why?

  1. To theoretical achieve a “data rate” of 300Mbps you need to have 40MHz wide channels, clients are 2×2:2 802.11n with an RSSI -64dBm using a short guard interval.
  2. The term “data rate” is completely miss leading this is not your actually throughput, wireless has a lot of overheads, basically cut that value in half and there is your starting point.

Now let’s look at some other variables that also affect the capacity of our wireless networks

  1. Business requirements:
    1. Different applications have different channel utilisation threshold before services start being affected
    2. If Voice services are required what are the RF design and configuration requirements,
      1. RF design
      2. Vendor specific configuration requirements: i.e. Cisco CAC
  2. QoS design
    1. Such as 802.11e, WMM, DSCP & CoS
  3. Rate limiting
    1. Can have significant impact to air-time utilisation
  4. Airtime fairness
  5. # of SSIDs
  6. % of associated users or concurrent clients?
  7. AP’s configuration i.e. 1x 2.4GHz and 1x 5GHz
  8. AP Chips sets, can they handle all the con-current and client associations
  9. Sources of RF and non-RF interference
  10. Wireless design requirements: Mesh with client connect, etc.
    1. Mesh present a whole number of extra requirements (headaches) when considering overall aggregate throughput per client.
    2. Channel plan
    3. Suitable Channel width without increasing CCI & ACI
  11. Device types, density, & number of devices:  A killer especially with legacy devices.
  12. Client device drivers are up to date.
  13. Backhaul: extremely important.

While an AP data sheet may say it can support # of clients per radio, we as wireless engineers need to ensure our wireless networks are designed with sufficient AP density to accommodate the numbers/types of clients and their throughput requirements.

Capacity planning is critical step to ensuring the wireless network meets the requirements of the design.

CCA from Another Perspective

As we know 802.11 is half-duplex, so devices must check the wireless medium to determine if it is free or busy before it can send traffic, No issues there.

Where it gets interesting is when we look at the thresholds (at a high level) that determine if the medium is busy. Part of the mechanism used to determine if it is available is called Clear Channel Assessment (CCA). Within CCA there are two thresholds – Signal Detect (SD)and Energy Detect (ED).

–          Signal detect (SD): listen for any transmitting 802.11 frames

–          Energy detect (ED): list for any other none RF transmissions that is 20dB stronger than SD (won’t go into this in this post)

SD is used to detect a transmitting 802.11 preamble, which is contained within the physical layer header of an 802.11 frame, if it can decode the preamble, it will attempt to sync with the incoming transmission, which is sent at the lowest basic rate for the Band, 1 Mbps for 2.4GHz and 6Mbps for 5GHz. The SD threshold is usual set at 4dB SNR, so if the noise floor is at -96dBm, an 802.11 device is able to demodulate the preamble at an RSSI of -92dBm, which would then trigger CCA, causing devices to deferrer.

So how far is -92dBm or 4 dB SNR? The below simulation shows you how far, represented by a simulated Cisco 3802 AP with the following configuration: 1mW, 2.4GHz and no obstruction,

The colour coding represents the following signal areas,

Screen Shot 2019-03-21 at 7.34.00 pm.png: The wireless coverage we have designed for, where we want our clients to be associated

Screen Shot 2019-03-21 at 7.34.06 pm.png : The area that can cause CCA

and every where else is where it becomes noise below the required CCA threshold

Screen Shot 2019-03-21 at 7.37.22 pm.png

  • Note the picture above is only the Radius.

Using 1mW (equates to 0dBm) and with a cell edged measure at -92dBm the total distance is 500m.

In some of the environments that I work in our industrial wireless AP can vary anywhere from 2Watts (33dBm) – 4Watts (36dBm) in the 2 .4GHz and 5GHz band and our enterprise AP around 25mW (14dBm) +/-, so the potential coverage area for SD to cause devices to deferrer can be quite large

Understanding this is important when it comes to our wireless designs. We design for a required cell edge i.e. -67dBm/25dB SNR for devices, but we also need to account for the signal we don’t care about as it continues on, so we can minimise Co- channel interference (CCI).

Minimising CCI is possible in 5GHz (depending on AP capacity), but not possible in 2.4GHz. It becomes an even greater pain in the butt when we add client STA’s into the mix, as the client STA are one of the biggest contributors to CCI.

Screen Shot 2019-03-21 at 7.38.27 pm.png

While we may have done our best to minimise AP CCI, as soon as a mobile client roams to the outer edges of a cell, it has the potential to extend the contention domain and interfere with another AP on the same channel, causing devices to defer. This is due to SD as mention previously

This in turn increases medium contention, due to devices having to defer, which eats up your airtime – equaling less throughput.

What can we do to help reduce this?

  • Understand your AP and client STA receive sensitivities levels
  • Consider your CCI boundaries in your designs
  • Use directional antennas where possible

Would like to give a shout out to two legends @WazzFi and @Stephen_Cooper for their input and support.

Reference:

Tom Carpenter (2016). Certified Wireless Analysis Professional (CWAP-402). US: CertiTrek Publishing

https://blog.aerohive.com/what-is-clear-channel-assessment-cca/

Devin Akin (2018) Certified Wireless Design Professional Training course