Help Center/ Content Delivery Network/ FAQs/ Troubleshooting/ Why Are CDN PoPs Offline During O&M?
Updated on 2024-08-15 GMT+08:00

Why Are CDN PoPs Offline During O&M?

Huawei Cloud CDN has over 2,800 PoPs in more than 130 countries and regions. To ensure the normal running of these PoPs, the O&M team of Huawei Cloud CDN irregularly performs O&M, such as software/hardware upgrade, network cutover, and PoP reconstruction. Generally, relevant PoPs are brought offline during O&M. Huawei Cloud CDN will take measures to reduce the O&M impact on users. However, due to the local carrier and client DNS caches, some users may still access offline PoPs, resulting in access failures.

Optimization Measures Taken by Huawei Cloud CDN

Routine PoP O&M is a normal maintenance and upgrade behavior, complying with the service availability commitment in the Content Delivery Network (CDN) Service Level Agreement. To minimize the impact on client access, Huawei Cloud CDN provides measures during PoP O&M, including but not limited to the following:

  • Very small number of concurrent PoPs in O&M
  • O&M time during off-peak hours in the early morning, rather than peak hours (19:00–23:00)
  • Traffic scheduling from PoPs in O&M to peripheral normal PoPs before PoPs are brought offline based on the standard PoP O&M process
  • Monitoring of exceptions (such as carrier line faults and data center faults) that may bring Huawei Cloud CDN PoPs offline unexpectedly. The scheduling system of CDN detects and monitors all PoPs. Once a PoP goes offline abnormally, the scheduling system schedules the traffic from this PoP to peripheral PoPs that are running properly in real time.

Optimization Measures on the User Side

Although Huawei Cloud CDN has taken measures to reduce the impact of O&M operations on clients, some clients still send requests to offline PoPs because both the local carrier and clients have DNS caches. As a result, the access fails. It is recommended that the following measures be taken on the user side to reduce the impact:

  • Clear the local carrier's DNS cache so that clients can obtain the latest resolution result for subsequent requests, preventing the resolution to offline PoPs.
  • Clear the client cache promptly. After that, when the client user accesses a PoP, the CDN scheduling system allocates the nearest available PoP.
  • Use the HTTPDNS technology. You can submit a service ticket to request the HTTPDNS scheduling function. You need to integrate the HTTPDNS technology on clients. Clients can bypass the local DNS of carriers and directly access the scheduling system of Huawei Cloud CDN using HTTPS. Then CDN schedules the optimal PoPs of domain names to be accessed. In this way, the PoP switchover latency caused by the local DNS cache can be avoided.