Experience in performance optimization and Performance Optimization

Experience in performance optimization and Performance Optimization

 

Communicate with colleagues. After a year or two of the previous online system, there were 20 millions or 30 millions million data records.

 

System, there is no way to deal with so much data

 

There are two reasons: one is that the data volume is not considered during design.

 

Another problem is that the data volume is not noticed when writing code.

 

Not to focus on the data volume, this may be the programmer's own consciousness, or it may be the customer's self-thinking not to pay attention

 

The customer said that we don't have much data, so you don't have to consider that much. It's hard to say that this system will not be used in the future.

 

Now, we only need to store 1 million pieces of bottom data, and 5000 pieces of transaction data can be run out.

 

It sounds like there are no more than 1 million basic data entries. It should be less difficult.

 

What the customer said is the same as what the customer did not.

 

Because at the beginning, there won't be 1 million of the underlying data, and there won't be exactly 1 million entries in five thousand, so we can process them.

 

1 million pieces of data is just a point. After this point is reached, what should we do?

 

In addition to poor performance bottlenecks in design, some problems can still be solved when writing code.

 

The following is my experience on optimizing the customer's sales report.

 

First of all, we should cultivate our own optimization meaning. When writing code, we should pay attention to this problem.

The other is that there is no problem to ensure that the customer's needs are prioritized. In this case, optimization will be performed if there is a performance problem.

If you do not consider whether the function is developed, you will be eager to optimize it. Some bugs may have been detected and need to be modified. After the modification, you will find that the previous optimization is good, optimization is required, which may waste a lot of time.

 

1. Optimization to customer satisfaction, but performance issues, the customer's requirements, we are not necessarily able to do

2. give full play to the computing power of the computer, so that the CPU and memory utilization can reach the extreme, reducing the number of interactions with IO, databases, and other systems

Quantitative analysis using performance testing tools

1. DotTreace is used to analyze the performance bottleneck of a program. You can see the specific execution time of each method and optimize it in a certain direction.

1. Ask experienced people for advice, who can solve problems more quickly. experienced people may give more reasonable opinions.