How fast does the burst bucket refill when the datarate is not maxed out at its cap?
Is there a way to change the sustained datarate cap without resetting the SM so its value could be adjusted based on total network traffic?
The burst bucket will refill at the MIR rate minus whatever is being used at that time. For example,
on a 4 Mpbs MIR radio, if 500kbps is being used, then the burst bucket will be filled at 3.5 Mbps.
There is a whitepaper that discusses this feature in detail here: Max Burst Whitepaper
Right now, there is not a way to make these changes without a reboot.
1 Like
That sounds confusing to me. I'll paraphrase, and let me know if I'm mistaken.
Let's assume the following:
4000 kbps sustained data Rate
8000 kbps max burst data rate
32000 kb burst allocation
While there is burst allocation left, it will pass data at 8 Mbps. The burst allocation refills at 4 Mbps, so once it's runs out the SM will throttle down to 4 Mbps.
The amount of time it takes for the burst to run out is (burst allocation) / (max burst rate - sustained burst rate), or in this case 8 seconds.
The allocation is the size of the bucket... 32 Mb in your case.
If the radio requires a lot of data, it will burst to 8 Mpbs for 4 seconds (4 s * 8 Mbps = 32 Mb)
After 4 seconds, it will reduce throughput to a sustained rate of 4 Mbps.
Once the link requires less than 4 Mbps, the bucket will refill. The rate of refill is the difference between 4 Mbps and what's being used at the time.
That is, if the link is not using data at all, it will refill that 32 Mb at a rate of 4 Mbps (i.e. it will take 8s to refill).
Did you read the whitepaper?
Sorry, Matt, you have a bit of an error in your calculation. :)
The burst bucket is filled at the MIR rate so in this case, every second it is getting up to 4Mbits added (if there is room in the bucket).
Hence, the actual consumption of the bucket bits is really the delta between achieved data rate (for this example, say max burst rate of 8Mbps) and MIR (4Mbps), so, 32Mbits are consumed at 4Mbps becuase the other 4Mbps are being replenished by the MIR.
So, in summary, Ted's example is correct - it will take 8 seconds (32Mbits / 4Mbps) to deplete the bucket bits, at which point, every second is just MIR in, MIR out, so that's the "throttle" to 4Mbps rate allowed.
When the usage drops down below 4Mbps, then the MIR fill every second will leave some unused bits in the bucket until the bucket fills back up to 32Mbits. If the rate goes to 0, then that will be 8 seconds at 4Mbps to get to 32Mbits back into the bucket.
It can get confusing, but its all about deltas and achieved data rates. The white paper does a good job of explaning this with helpful graphics to make it easier to understand.
1 Like
Thanks for setting me straight Aaron...