When is there a CPU or Main Memory Bottleneck?

最新推荐文章于 2025-08-21 14:18:22 发布
原创 最新推荐文章于 2025-08-21 14:18:22 发布 · 392 阅读 · 0 ·
CC 4.0 BY-SA版权
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。

本文探讨了系统性能优化的关键指标，包括闲置CPU容量、页面速率以及硬件瓶颈的常见原因。针对不同类型的计算机配置，提供了内存与虚拟内存比例的指导原则，并详细解析了识别系统中关键进程的方法。
The unused CPU capacity (idle) should normally average at least 20% per hour. This enables the system to accommodate temporary workload peaks. A reading of 35% idle CPU capacity is even better. For the paging rate, the following guideline values apply:


->For computers that contain a database, an SAP J2EE Engine, or TREX, only very minor paging rates should occur; that is, they should be dimensioned in such a way that the available main memory corresponds to the configured memory areas.


->For computers that only include ABAP instances, you can tolerate moderate paging rates of up to 20% of the physical main memory per hour.


For instance,Main memory bottleneck, As a rule of thumb, there should be approximately 50% more virtual memory than physical memory.


Causes of Hardware Bottlenecks

->Incorrect load distribution (It is extremely important that the database server has enough resources.)


->CPU load of individual programs (In the operating system monitor (Transaction ST06), select the analysis SNAPSHOT • TOP CPU PROCESSES.)

* You can identify the following processes:

--> dw_<instance>:SAP work process of the ABAP Server on a UNIX operating system. On Windows operating systems, the name is “disp+work.”

-->jstart ...                :Server process of SAP Java instance.

-->TREX ...               :TREX process. The server type is indicated in the process name, for instance, index server, preprocessor, and so on.

-->icman ...              :Process of the Internet Communication Manager (ICM).

-->saposcol ...         :Auxiliary program, which collects the data for the operating system monitor





A Survey on Deep Learning Techniques Applied to medical image analysis
AI天才研究院
08-13 645
作者：禅与计算机程序设计艺术 Introduction:Deep learning techniques have recently gained popularity in medical image analysis because they are capable of accurately identifying disease markers without relying on human intervention, enabling better decision-making and tre
【模型精调LoRA】LoRA 低秩适应微调的工作原理和代码实现示例 What is LoRA? Low-Rank Adaptation for finetuning LLMs EXPLAINED
AI天才研究院
03-11 1835
LoRA 是一种有效的大模型微调技术，可以提高推理效率并保持良好的性能。LoRA 在许多下游任务中都取得了良好的效果，包括文本分类、机器翻译和问答。
Detecting Memory Bottlenecks （检测内存瓶颈）
fyifei0558的专栏
03-04 636
Detecting Memory Bottlenecks http://technet.microsoft.com/en-us/library/cc749872.aspx The Virtual Memory Manager moves pages of code and data between disk and memory in a process called pa
SQL Server 2008性能故障排查（一）——概论
www.v5qq.com的技术博客
07-05 457
备注：本人花了大量下班时间翻译，绝无抄袭，允许转载，但请注明出处。由于篇幅长，无法一篇博文全部说完，同时也没那么快全部翻译完，所以按章节发布。由于本人水平有限，翻译结果肯定存在问题，为了不造成误导，在每篇结尾处都附上原文，供大家参考，也希望能指出我的问题，以便改进。谢谢。 另外，本文写给稍微有经验的数据库开发人员或者DBA看，初学者可能会看不懂。在此请见谅 作者：Sunil Agarw...
SQL Server 2008性能故障排查（三）——I/O
www.v5qq.com的技术博客
07-15 441
接着上一章：CPU瓶颈 I/O瓶颈（I/O Bottlenecks）: SQLServer的性能严重依赖I/O子系统。除非你的数据库完全加载到物理内存中，否则SQLServer会不断地把数据库文件从缓存池中搬进搬出，这会引起大量的I/O传输。同样地，日志记录在事务被声明为已提交前必须写入磁盘。最后，SQLServer基于许多原因使用tempdb，比如存储临时结果、排序和保持行版本。所以一个...
RNNS ARE NOT TRANSFORMERS (YET): The Key Bottleneck on In-context Retrieval 翻译
Doc2X的博客
11-22 1022
https://arxiv.org/pdf/2402.18510 RNNS ARE NOT TRANSFORMERS (YET): The Key Bottleneck on In-context Retrieval 翻译 RNNS 还不是 TRANSFORMERS（目前）：上下文检索的关键瓶颈
Real-Time Rendering——18.2 Locating the Bottleneck找出瓶颈
m0_37609239的博客
10-17 269
Real-Time Rendering——18.2 Locating the Bottleneck找出瓶颈
CPU Utilization is Wrong
u013597671的专栏
05-30 801
The metric we all use for CPU utilization is deeply misleading, and getting worse every year. What is CPU utilization? How busy your processors are? No, that's not what it measures. Yes, I'm talking a
什么是“缓存友好”代码？
asdfgh0077的博客
04-13 1809
What is the difference between " cache unfriendly code " and the " cache friendly " code? “ 缓存不友好的代
What is the Python Global Interpreter Lock (GIL)?
bbbeoy的专栏
12-06 332
Table of Contents What problem did the GIL solve for Python? Why was the GIL chosen as the solution? The impact on multi-threaded Python programs Why hasn’t the GIL been removed yet? Why wasn’t i...
PAMI论文研读-JARV IS-1: Open-world Multi-task Agents with Memory-Augmented Multimodal Language Models
最新发布
优快云XXCQ的博客
08-21 790
具体来说，我们在预先训练的多模态语言模型之上开发JARV IS-1，该模型将视觉观察和文本指令映射到计划。计划将最终发送到目标条件控制器。我们为JARV IS-1配备了多模态记忆，这有助于使用预先训练的知识和实际游戏生存经验进行规划。
The MySQL 8.0.0 Milestone Release is available
帅的数说
09-20 2152
http://mysqlserverteam.com/the-mysql-8-0-0-milestone-release-is-available/ The MySQL Development team is happy to announce our 8.0.0 development milestone release (DMR), now available for downloa
The Performance Monitor in the DBA Cockpit
scmaxyes的专栏
01-14 1025
The database monitor in SAP NetWeaver AS ABAP collects performance data from two sources: ->The monitor relies on performance data collected by the relevant database system.  -> A portion of the per
Identifying Expensive SQL Statements
scmaxyes的专栏
01-14 479
Expensive SQL statements are long-running statements and one of the main causes of performance problems. Expensive SQL statements can have the following effects on the entire system: ->They ca
Analyzing the Database Buffer
scmaxyes的专栏
01-14 452
The following values characterize the quality of data buffer accesses: -> Physical read accesses Number of read accesses to the hard drive. This value indicates how many blocks or pages must be load
Detailed Analysis of a Hardware Bottleneck
scmaxyes的专栏
01-14 447
Performance problems may be indicated if: -> The average idle CPU capacity is less than 20% every hour. -> More than 20% of the physical main memory is paged every hour. -> Utilization of individua
Database Monitoring
scmaxyes的专栏
01-14 429
The most important functions that you require for performance monitoring of the database include the following: ->Overview of the database buffers' status ->Overview of the currently running databas
Identifying Read/Write (I/O) Problems - Database
scmaxyes的专栏
01-14 391
To achieve optimal database performance, 110 activity (read/write accesses) should be evenly distributed on the database's hard drives. You should avoid hot spots. You can find important key figures
there is very likely to create a memory leak
07-18
### Identifying and Preventing Memory Leaks in Software Development Memory leaks are a common issue in software development, particularly in applications that manage memory manually. A memory leak occurs when a program allocates memory and fails to release it after it is no longer needed. Over time, this can lead to excessive memory consumption, degraded performance, or even application crashes. Detecting and preventing memory leaks is crucial for maintaining application stability and efficiency. #### Identifying Memory Leaks To identify memory leaks, developers can use a combination of tools and techniques: - **Profiling Tools**: Tools like Valgrind for C/C++, VisualVM for Java, or Instruments for macOS and iOS can track memory allocations and detect memory leaks. These tools provide detailed reports on memory usage over time and can pinpoint areas where memory is not being properly released. - **Logging and Debugging**: By logging memory allocation and deallocation events, developers can manually inspect logs to identify discrepancies. Setting breakpoints in a debugger can also help track down where memory is being allocated without being freed. - **Unit Testing**: Writing unit tests that monitor memory usage can help detect memory leaks in specific functions or modules. For example, running a function repeatedly and observing if memory usage increases linearly can indicate a memory leak. - **Code Reviews**: Regular code reviews can help catch memory management issues early. Reviewers should look for patterns such as missing `free()` calls in C, unbalanced `retain`/`release` in Objective-C, or failure to close resources in languages like Java. #### Preventing Memory Leaks Preventing memory leaks involves adopting best practices and using modern programming techniques: - **Use Smart Pointers (C++)**: In C++, smart pointers like `std::unique_ptr` and `std::shared_ptr` automatically manage memory, reducing the risk of leaks. These pointers ensure that memory is released when it is no longer referenced. - **Garbage Collection (Java, .NET)**: Languages with built-in garbage collection, such as Java and C#, reduce the burden of manual memory management. However, developers should still be cautious with object lifetimes and avoid unintentional object retention, such as holding references in static collections. - **Resource Management Patterns**: Use patterns like RAII (Resource Acquisition Is Initialization) in C++ or try-with-resources in Java to ensure that resources are properly released when they are no longer needed. - **Avoid Circular References**: In environments with reference counting (e.g., Objective-C, Python), circular references can prevent objects from being deallocated. Using weak references can help break these cycles. - **Use Memory-Safe Languages**: Consider using memory-safe languages like Rust, which enforce memory safety at compile time without relying on a garbage collector. #### Example: Detecting a Memory Leak in C ```c #include <stdlib.h> #include <stdio.h> void leak_memory() { char *buffer = (char *)malloc(100); // Forgot to free the buffer } int main() { while (1) { leak_memory(); sleep(1); // Simulate ongoing process } return 0; } ``` In this example, the `leak_memory` function allocates memory but never frees it. Over time, this will cause the program's memory usage to grow indefinitely. Tools like Valgrind can detect this leak and report the source. #### Example: Using Smart Pointers in C++ ```cpp #include <memory> #include <iostream> void safe_memory() { std::unique_ptr<int> ptr(new int(42)); std::cout << *ptr << std::endl; // Memory is automatically freed when ptr goes out of scope } int main() { safe_memory(); return 0; } ``` Here, `std::unique_ptr` ensures that the allocated memory is automatically released when the pointer goes out of scope, preventing a memory leak. #### Example: Using Try-With-Resources in Java ```java import java.io.*; public class ResourceExample { public static void main(String[] args) { try (BufferedReader br = new BufferedReader(new FileReader("file.txt"))) { String line; while ((line = br.readLine()) != null) { System.out.println(line); } } catch (IOException e) { e.printStackTrace(); } } } ``` In this Java example, the `BufferedReader` is declared within a try-with-resources statement, ensuring that it is closed automatically at the end of the block. ###
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值