Top 10 classic network O & M challenges of data centers

Top 10 classic network O & M challenges of data centers

As data center construction grows and new technologies increase, the network carrying data center services becomes extremely complex. To adapt to the development of data center services, the data center network is constantly updated and changed, which makes O & M very difficult in the future. Network O & M of data centers is a common issue and the most prominent issue in data centers. This is mainly due to the closeness and arrangement of network technologies, of course, there is also a relationship between the network protocol and the complexity of device design, which makes it much more difficult to master the network O & M essentials than to master other technologies, therefore, various problems occur during network O & M. Once a network device encounters a problem, especially a core network problem, the entire data center business will be affected, sometimes there is no network backup available. In such a crisis, the comprehensive troubleshooting capability of O & M personnel is tested. Various factors have caused various problems in network O & M, this article will talk about these difficulties to see if you share the same feelings in data center O & M and whether there are better countermeasures.

Problem 1: too many manual operations are expected to be reduced

Network O & M personnel in the data center are most afraid of network changes. because too many command operations are involved, errors may occur if the operations are poor. If the network O & M personnel can be automatically deployed, the O & M personnel's work time can be greatly reduced and it is not prone to errors, as the network O & M personnel of the data center, there is no need to have too much knowledge about the underlying network commands, as long as the network changes meet business needs. In fact, this type of problem is the most prominent in O & M, and many network equipment commands are obscure. O & M personnel do not have the time and ability to read every RFC document, what is needed is a simple and clear solution. The emergence of SDN may reduce the dependency of O & M personnel on manual operations. However, it is still unknown to what extent SDN can develop in the future.

Problem 2: it is difficult for network changes to keep up with requirements.

The requirements of data center business departments are diverse, especially for performance. Many unreasonable demands are also accepted, and difficulties are discovered only when implementation is completed. Many business departments do not have a clear understanding of the data center network or what the existing network can provide, which leads to a gap between the two sides, in the end, many requirements cannot be implemented through network changes, or network changes may affect existing services, causing a high price.

Problem 3: collaboration between network operations and system integrators

The network is only the most important part of the data center. Any service operation is inseparable from the network. Therefore, any operation on the network must communicate well with other system modules to avoid affecting the operation of the entire system. This involves dealing with system integrators.

Problem 4: busy with maintenance, difficult to quickly deploy new businesses

If a data center network is designed with inherent defects, frequent problems are inevitable. Such data center network O & M personnel are also busy dealing with various network problems every day, especially the problems that have affected business operation, so they have no energy to deploy new businesses. Such a vicious circle leads to the failure to promote the entire data center's business and eventually the loss of a large number of customers.

Problem 5: network deployment troubles

Devices in the data center must have their own IP addresses or MAC addresses to achieve interconnection. These are used to represent their identities in the network. O & M personnel must adapt to these identities in the network, such as distributing dynamic routing learning or static routing, and need to configure gateways and DHCP, these configurations even need to be deployed on all network devices across the network. Some data centers have hundreds of network devices from the core to the access, and one configuration is obviously very troublesome. How to reduce the workload is particularly meaningful for improving the O & M efficiency.

Challenge 6: using simple tools to manually manage IP addresses

Network O & M personnel usually need to manage the IP addresses of these devices, so that they can find the desired IP address when using or failing. This quantity is massive. It is common to have tens of thousands of servers in a large data center. It takes a long time to sort out the IP addresses of these devices. O & M personnel can only manage the data through a simple Excel table. They can search for the data and record the data when there is an update. The data must be updated in real time to be accurate, this requires O & M personnel to invest a lot of effort to maintain this form, which is cumbersome.

Challenge 7: A wide range of network devices make it difficult to fully master

The biggest headache for O & M personnel is the wide variety of network devices. Different manufacturers have different command styles and meanings, and even different models of devices of a single manufacturer are different. This makes network O & M extremely difficult. O & M personnel have to master the basic operation commands of all devices in the data center and spend a lot of time familiarizing themselves with these devices, generally, there are thousands of network device commands, and it is basically impossible to fully master them. In addition, the O & M personnel will go crazy when different models of devices are used.

Problem 8: the technical level of the network management team is not high

Currently, the network management system of the data center monitors the running network devices, but in fact it extracts the logs and alarms from the devices, and then provides some alarms, in addition, some device information can be obtained through the network management. In fact, network management does not support O & M very much. Real smart network management should replace some of the work of O & M personnel, such as distributing configuration changes, switching networks for business failures, and network self-check. Through network management, real intelligent network management should be achieved, to reduce the workload of O & M personnel, the network management technology needs to be further improved.

Problem 9: too many tools to master

There are more than 8000 Ethernet RFC protocols, according to the five layers of the network, there are a variety of protocol definitions. It is the diversity of network protocols. Therefore, we need to design many auxiliary tools to master it and use many tools for network analysis. For example, XPING, Tracert, packet capture tools, IP mask conversion, and so on. There are many of these tools, many of which are open-source tools on the network. There are various bugs that make it inconvenient to use, however, you have to use it during network O & M. Sometimes you can develop a small tool if you cannot use it. Because of this, so many network analysis tools appear on the network.

Problem 10: Hard O & M and Low Income

Network O & M is a function with low cost performance. As an important part of the data center, the importance of the network does not reflect the revenue of network O & M, as a result, no one is willing to conduct in-depth research on the O & M work. Most O & M personnel work 1 ~ 3 years of junior technical staff, lack of senior network experts for more than 10 years, which makes the O & M level of the data center unable to be improved.

Obviously, the network O & M of the data center faces many difficulties and is a short board in the data center. Which data center can solve the problem of network O & M, so it can be well mixed in this circle.