VM Summary information is more difficult to obtain than the previous host, memory, cluster, because in the performance side need to do analysis. And the most valuable of these VMS is the performance analysis and which resource pool belongs to. The rest of the summary is basically no difficulty. And there are a series of data that needs to be converted, such as memory, the data that the hard drive gets is divided by 1024来, but the CPU value is 1000来 converted. So it takes a lot of attention to get the basic data.
Our company currently only needs to obtain data, rather than using Java to control the operation of the CDH client. The final task is also falling in the performance of the analysis. The real-time monitoring rule for VM machines is 20/s scan once. Specific to day/week/year, is obtained by means of the average. For example, for a day of VM performance, the day's data can not get 20S sweep out of the value, but the system calculates the average, and then recorded to a certain time of day. I'm doing every five minutes of the current value, the program runs for a day, and then compared to the value of the VM on the day of CDH, the discovery is not the same, because I am getting the current value, but the value of CDH is a series of average records, then you need to tell the boss why the data is inconsistent.
Real-time performance acquisition here, each scan will generate 180 records of the near time, I am taking the peak of a value as the most recent value. Specific nest in the provision of code reference it.
VMware's VM Machine