Two unconventional solutions for server slowdown

For websites, traffic is the pursuit, but for webmasters, server speed is fundamental. No Webmaster can tolerate server slowdown. As a server maintainer, I am very anxious when the server slows down.

From the two recent attempts to solve server slowdown problems, it is sometimes difficult to think that server slowdown will be caused by these reasons. I hope you will think of this idea when your server slows down one day!

Server slowdown unconventional solution 1

Scenario:
A server is hosted in the Netcom data center, which is exclusive to MB. I spent a lot of money, of course, to create greater value. This server is mainly used for Vod. However, from the date of hosting, we never felt how fast M exclusive exists. After detection, we found that our company's 50 m optical fiber speed was not fast yet. So I wrote a speed test report and submitted it to China Netcom. I hope I can handle this issue very seriously.

After testing, the IDC room staff were very helpless. The network cable was inserted to other servers, and the speed was very high. The average download speed was 10 MB on the LAN. However, the average speed of downloading the same file on my server is only 6 MB, and the NIC is still the same. The conclusion is that the problem lies in my server.

The problem lies in my own server. I don't know how ignorant I am about the submitted test report. What I think is how to solve the speed problem? Are both NICs broken? In a helpless situation, we start to point this point in the system. Suddenly I found a place worth a try. That is, the connection speed of the NIC. Our current NICs are both 10 M/M Adaptive NICs. Generally, the connection speed of the NIC is not adjusted manually. However, in this helpless situation, only one-to-one tests can be performed. Select the full duplex mode of MB. After the server is restarted, test the speed again. After downloading the LAN file, the system returns to normal. In other places, the on-demand video is obviously faster.

(Note: The two images are not taken from the NIC attribute page of a server, but to illustrate the problem)

Server slowdown unconventional solution 2

Another server in our company does not know when it will become very slow. I consulted the data center and people said it was normal, maybe weProgram. So I began to study every program and the memory occupied by every website. However, after a long time of research and optimization, the effect is still the same, and it is still slow and intolerable.

At this time, I have been using the ping command to check the server speed. Since the server is slow, packets returned by Ping are always lost. In severe cases, one packet loss occurs on average once every 10 times. Careful analysis of packet loss should not be a program problem, but a line problem, ping our server from other places is still causing serious packet loss, which eliminates the problem of our company's network. I thought again whether the NIC has been working for too long and is aging, so I asked the IDC staff to change the NIC again.

The funny result is that the IDC staff told me that after their tests, they found that the problem was caused by the network cable. They changed a new network cable and the speed was back to normal. There is no packet loss.

When your server has no problems, you may feel that the above two methods are ridiculous, but the server runs 24 hours a day, and it is impossible to avoid other problems such as hardware aging, I only hope that my solution can become a way of thinking for your daily post-processing server.