I’m using 2GHz-5Mhz band. If I’m giving all my clints 2Mbps download/256Kbps upload, and they are at full blast, how many clients can I have without noticing much degradation at client’s end? I have aggregated 40M pipe to my ISP. Example calculation would be appreciated so I can learn.
In Simple Queues, if I allocate 512K/1M upload/download max limit for a group, is everyone in that group combined usage is limited or is individuals in that group limited? In other words, if there are two IP in a group, are each IP limited to 512K/1M (altimately 1M/2M) or both IPs combined total is limited to 512K/1M?
I’ve searched everywhere but I can’t find these basic tutorials… Thanks.
First question: if your clients have a full charge of load then 40M are used for max 20 clients without degradation, beyond this may occur slow downs for all. I’m not sure how to reach this consumption in your wireless schema, because multiple factors may affect the service in a wireless links. But note 20 clients at full charge is a dramatic scenario. With certain rules in queueing you should multiplex the total bandwidth of your provider between ten and twenty times the bandwitdh that you have sold. In example: 40M x 10 = 400M / 2M (by each client) = 200 clients with a bandwidth NOT guarantee of 2M each… note!!: not guaranteed but with a very good service (close to 2M the major part of the time) along the day.
Second question: In simple queues if there are two IP in a group both IPs combined total is limited to 512K/1M (not per client). If you want to limit each ip to 512k/1M you must use queues trees.
Isn’t the 2GHz-5MHz has maximum of 13MHz bandwidth? If so, at 2MHz full blast per client, wouldn’t it be limit to about 6-7 clients, even if I have 40M fiber connection?
Thanks for clarifying the queue, it was confusing me. One other question, does the queue take priority over AP/Client TX Limit in Access List setting?
Isn’t the 2GHz-5MHz has maximum of 13MHz bandwidth? If so, at 2MHz full blast per client, wouldn’t it be limit to about 6-7 clients, even if I have 40M fiber connection?
Don’t be confused between MHZ and MBPS. The transmission of IP packets is NOT continue, this fact improves the usage of the frequency. Since at IP level the communications betweenn AP and CPE is a ultra faster sequence of bidirectionals packets, this happens in a faster way in a queue format. The AP assign different time “windows” to each client to transmit in concurrent connections, in this way the ap optimizes her resources to be more efficient with al CPE connected to him. More clients depending of the same AP, then more wait time has the CPEs for “talk” to the AP. The concurrency is a question of how much CPEs are connected (and transmitting) at the same time to the same AP. Obviusly when major is the mhz used in a channel, more capacity have you network. Anyway you must test your scenario in the field: there are the best practice to learn how good work your devices. Try all.
One other question, does the queue take priority over AP/Client TX Limit in Access List setting?
Between AP and CPE the tx limit in access list have priority over queues because the simple queues or the pcq trees act when the packets has joined the router, the CPE / AP communications occurs in a previous instance.
Thanks for clarifying the priorities. I’m trying to get much info possible to resolve some issue I’m facing…
If I have 10 CPE’s and bandwidth equally assigned 1Mbps each, does the AP assigned the time “windows” equally to all CPE’s or based on needs? Can AP assign more time “window” to one CPE causing other CPE’s to frequently timeout?
No. The time is limited to bytes (not to msecs). The max limit is one frame of 2048 bytes, over there the AP assign the “window” to the next CPE. More charge of traffic, then more easy is reach the 2048 bytes frame limit and more time to complete the request is necessary. If the traffic downs below 2048 bytes by frame (i.e. low traffic), then the next “window” comes quickly and, of course, the AP performance is better.
Other things to note is the connection rate. If the clients all are linked at 54meg then the max possible transmission on a single AP is around 7.5 meg. If clients are linked at lower rates, this will reduce your overall AP thoughput. Assuming that you have ALL of your clents at 54meg only, then figure around 20-25 clients per AP given that they all use BW on average, giving you still around 7meg total thoughput. That assumes a decent balance between business and residential customers to get a 6:1 oversubscription rate. Some people would think that is high and some would think it is low, but it really depends on your customers.
Given that data, if you havfe 5-6 APs, you should support between 100 and 150 subs with 2 meg packages.
On your second question, I think it was answered, but you can use a single queue per IP, or PCQ, but you don’t have to use a Queue Tree if you don’t want.
