I was (am still) looking to do some math for taking total bytes/sec from perfmon readings and converting into percent utilization, and just realized that the multiplier= parameter works for division when it is given decimal number value.
For example:
Full utilization on a half-duplex 100 Mbs pipe would be 100,000,000 (roughly) or 200,000,000 on a full-duplex link. However, Ethernet is never 100% utilized due to overhead and pretty much caps out at 80% even on a good switch. So to start with, 100% utilization is really about 80,000,000.
Our counter is using megabytes so we are able to convert to megabits with a straight "multiply by 8", which turns the 10,mil readings into 80,mil readings. Voila, we are at our 100% mark.
The other key variable here is time (megabytes per seconds from perfmon, and megabits per second from bandwidth). However we don't have to mess with time any because it is constant for all variables.
Now we just have to bring it into percentage. Since we are already dealing with base-10 numbers ("100 megabits" is 100 which is base-10) we just have to scale this down, so a simple "multiplier=0.000001" will bring the 10s of millions down to 10s right away, and that gives us exactly what we are looking for--percent utilization for Mbit/s.
With different speed networks you need a different scale, of course. You will have to use one less decimal for 10 megabit links for example.
Things are a little more tricky with full-duplex and very-high-speed links (nothing hardly fills gigabit pipe so you have to find the natural cap and modify accordingly). In my case I notice that I am only able to get about 120 megabit/sec out of my full-duplex server link, so I have set the base multiplier to 0.000008 here. Full-speed uni-directional transfers that hit 80% half-duplex utilization (the normal) show about 60% of the bar graph, while the max full-speed utilization of 120% gets right up to the 100% mark without hitting it.
For example:
Full utilization on a half-duplex 100 Mbs pipe would be 100,000,000 (roughly) or 200,000,000 on a full-duplex link. However, Ethernet is never 100% utilized due to overhead and pretty much caps out at 80% even on a good switch. So to start with, 100% utilization is really about 80,000,000.
Our counter is using megabytes so we are able to convert to megabits with a straight "multiply by 8", which turns the 10,mil readings into 80,mil readings. Voila, we are at our 100% mark.
The other key variable here is time (megabytes per seconds from perfmon, and megabits per second from bandwidth). However we don't have to mess with time any because it is constant for all variables.
Now we just have to bring it into percentage. Since we are already dealing with base-10 numbers ("100 megabits" is 100 which is base-10) we just have to scale this down, so a simple "multiplier=0.000001" will bring the 10s of millions down to 10s right away, and that gives us exactly what we are looking for--percent utilization for Mbit/s.
With different speed networks you need a different scale, of course. You will have to use one less decimal for 10 megabit links for example.
Things are a little more tricky with full-duplex and very-high-speed links (nothing hardly fills gigabit pipe so you have to find the natural cap and modify accordingly). In my case I notice that I am only able to get about 120 megabit/sec out of my full-duplex server link, so I have set the base multiplier to 0.000008 here. Full-speed uni-directional transfers that hit 80% half-duplex utilization (the normal) show about 60% of the bar graph, while the max full-speed utilization of 120% gets right up to the 100% mark without hitting it.
Looking for additional LCD resources? Check out our LCD blog for the latest developments in LCD technology.