> For the complete documentation index, see [llms.txt](https://cooleditphoto.gitbook.io/muziqiu/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cooleditphoto.gitbook.io/muziqiu/arts-week2.md).

# ARTS WEEK2

1. I finished a classic two-pointer problem- minimum size subarray sum. The method of sliding window can apply to this problem perfectly.

leetcode 209

```java
   class Solution { public int minSubArrayLen(int s, int[] nums) {
    if(nums.length==0){ return 0; } 
    int left=0; 
    int right=0;
     int length=nums.length; 
     int result=length+1;
     int sum=0;

    while(right<length){
        while(sum<s&&right<length){
            sum+=nums[right++];
        }
        while(sum>=s){
            result=Math.min(result,right-left);
            sum -= nums[left++];
        }
    }

    if(result==length+1){
        return 0;
    }else{
        return result;
    }
}
```

2\. Review&#x20;

first part of Distributed Systems for Fun and Profit chapter 1

This chapter talks about some basic concepts of distributed systems, such as performance and availability.

`Distributed systems are constrained by two physical factors:`

* `the number of nodes (which increases with the required storage and computation capacity)`
* `the distance between nodes (information travels, at best, at the speed of light)`

`Working within those constraints:`

* `an increase in the number of independent nodes increases the probability of failure in a system (reducing availability and increasing administrative costs)`
* `an increase in the number of independent nodes may increase the need for communication between nodes (reducing performance as scale increases)`
* `an increase in geographic distance increases the minimum latency for communication between distant nodes (reducing performance for certain operations)`

At first glance,  I think more nodes bring more benefits rather than problems. I think as the traffic can be divided to different nodes, then the problem of computing(the traffic cost a lot of time for CPU) can be tackled. I just ignored the failure and the communication part of a distributed system!

3\.

tip:

how to solve bugs in an old system?

First, find all the classes and methods that are related to the bug.&#x20;

Second, read the code and the call method.

Third, add some logs and run the code.(mock or write a main method)

Finally, solve the bug!

4\. Share:

Kafka文档设计部分翻译

4.2 持久性

**不要害怕文件系统**

文件系统和网络差不多快

1高通量的文件读写比一般的disk seek快很多。可预测性强的使用方式。操作系统优化好。现代操作系统的策略是先读后写。大批量拿数据，小批量逻辑写。

2磁盘缓存大量用主存，如果不是直接i/o，即使是进程中缓存，在OS的页缓存。

3Kafka基于JVM,内存用量常是数据用量的两倍，堆内数据增长，垃圾收集变得困难。所以最终选择了文件系统，并认为pagecache优于内存缓存。我们至少因为自动获得全部空闲内存而获得了两倍空间，而存储压缩的比特结构而不存储一个个对象，可能用让我们得到了之前的两倍。

32GB的机器上，我们就能得到28-30GB的缓存，而且没有GC造成的损失。并且这个缓存能保持warm，即使服务重启，而进程内存则需要重新恢复（10GB月需要10min），或者缓存完全是冷的。而且这也简化了代码逻辑，因为这两者的coherence都归OS管。

最终设计是：所以数据立刻写到一个文件系统中的持久化log，不需要一次性的涌入文件系统。

这种以pagecache为中心的设计可以参见Varnish此文<https://varnish-cache.org/docs/trunk/phk/notes.html>。