Hi
I am doing an assessment for a customer who has asked us to evaluate their Meraki wireless setup and make a recommendation whether it has got enough capacity to accomodate all the wireless users they are expecting in a couple of months time. They are replacing desktops with laptops for everyone hence forseeing a lot more usage of wireless than before. While we wont be able to comment on the coverage aspect of it without the survey, how can we make a best possible recommendation based on the data we have got (expected number of users, info from dashboard)?
Linked to the same question, how can we know what is the possible throughput value for any AP in reality (not the one in datasheet as I know that is a theoretical figure)?
Thanks,
this document gives you a basic view of what application data takes how much BW and what are realistic wifi speeds. and how users per AP are calculated.
Thanks @ww for this. I saw this guide which does look comprehensive but I cannot understand how they got to the value of 101Mbps as AP's total throughput. That bit of the document is confusing.
If I want to do the same for MR 18 which is 802.11n with 2x2:2. Does that mean I come to the value of throughput 50Mbps (which is staggering compared to the value of 600Mbps shown in the datasheet)?
@ccie_ma the MR18 has two 2x2:2 radios and the 600Mbps is the combined raw data rate. The MR18 is also capable of using 40MHz channels and the document is for high density deployments where the use of 20MHz channels is common to avoid interfering with neighbouring APs. This does indeed lead to each radio having a usable throughput of ~50Mbps, so 100Mbps for the AP. I do also believe that the 600Mbps is only achieved with the 2.4GHz radio using a 40MHz channel that uses nearly all available 2.4GHz channels...
Modern APs do perform much better, but if I were the company I'd be buying docking stations and using the wired ethernet if possible. If that isn't possible buying a current WiFi6 AP model would help as the laptops should support this and you would get a lot better performance.
The MCS11 data rates of the MR56 are 1.1Gbps for 2.4GHz and 4.8Gbps for the 5GHz radio. Comparing this to the 300Mbps of each radio in the MR18 and limiting to 20Mhz gets you to 0.57Gbps and 1.1Gbps, then extrapolating the limitations and inefficiencies would lead to throughput of about 200Mbps for 2.4GHz and 400Mbps for 5Ghz which would be about right with testing I have done, though you may get about double of that due to the additional streams available which a single client cannot use.
I'd estimate that a single MR56 with 20Mhz channels would give you 600Mbps-1Gbps reliable throughput with 30 clients connected to each radio though the safer bet would be nearer the lower end of the range. Remember that other devices; phones, watches, printers etc. may well also connect to the WiFi and you want to size for no more than 30 devices per radio. More will work, but efficiency will be even worse.
Thanks @cmr . So shall we just take the number of users as a rough guide to calculate the number of users. i.e if best practice is not to allow more than 25-30 users per AP then we can use that as a limiting factor. Throughput of AP becomes less significant in calculations then. Do you agree?
> if best practice is not to allow more than 25-30 users per AP
You previously said the director wants to provide an "excellent" experience. I would aim for 20 users per AP given that specification.
Thanks Philip for your feedback - Would you say the number of users is per AP or per radio ? MR 18 for example has 2 radios. Would we say it can support total of 25+25 users ?
I do it per AP.
Even though the MR18 has two radios, it does not have anywhere near enough processing power to handle 50 users. It would provide horrible performance.
Wow, what a tough question without more context.
I tend to allow 1 AP per 20 users to give a good experience.
I agree with @cmr. As the users are currently using desktop machines on ethernet then using docking stations using the exisitng ethernet cable will save any potential bandwidth issues and it still means the users can take devices home and work on wifi if needed.
We recently replaced desktops for laptops in our finance department and they are all using docking stations mainly because of external monitors but also for ethernet because they are working from mapped drives and we wanted to reduce the risk of issues this may have caused on wifi.
Thanks @BlakeRichardson the customer director wants to provide excellent wireless service to the users. They might end up using docking stations for the reasons you mentioned (dual/big screens etc) but for the purpose of this assessment, we need to make sure their wireless deployment can serve all the potential users. So we have got the expceted number of users for each site and their current AP profile from the client and need to assess whether this AP profile (number and make) can cater for the no of users expected? (Ignoring the coverage aspect of it for this discussion which would be covered in survey separately)
Is the office a high density area? I would play it safe and when you find the desired model of access point choose the next model up which will give you some wiggle room with expansion.
One more thing that confuses me in this best practice guide is how they have calculated the number of APs based on the client count. They are assuming 30/70 % split for 2.4Ghz/5Ghz users but then they calculate the number of users just based on 5Ghz users. Why are they ignoring the 180 2.4Ghz users in this calculation?