Computer Room Collection: How to solve server cooling failure?

Network infrastructure design is a complex process. Every small decision may affect the server in the data center. The most overlooked is the server cooling demand. It is not a misjudgment or a serious underestimation. Everything happens in all kinds of environments. Historical experience has shown that tight IT budgets are forcing data centers to ignore critical cooling and temperature control needs, even in large data centers. Only by effectively preventing server cooling failures can the application running on the server never go offline, and the company can really save money. So how to do the cooling work of the server, how to deal with it when the fault occurs?

Server Cooling Capacity

Before dealing with failures in the server's cooling infrastructure, it is important to understand how the cooling system works and how much cooling capacity is needed, cooling and data center placement. The number of electrical equipment is closely related, and the expected power consumption can be calculated based on the amperage number marked on the equipment. The larger the amperage, the more cooling capacity is required. The easiest way to calculate is to know the power consumption of the server (per server or all servers). Usually, a single CPU server needs 1 amp or 120 watts (1 amp x 120 volts = 120 watts), dual CPU (such as Strong or AMD) servers need 2 amps or 240 watts, and finally multiply the required cooling capacity per watt (about 3.4 BTU) to calculate the amount of cooling the server needs.

For example, a 120 watt server requires 408 BTU of cooling capacity (120x3.4). Use this method to calculate the amount of cooling required by all servers and consult a certified HVAC consultant to see if the calculation is feasible. .

Handling server refrigeration system failures

When a server refrigeration system fails, the most important thing is to make quick decisions and take action. The correct action will make the system run longer. And prevent more serious failures from happening again. The following guidelines can be referenced in the event of a refrigeration system failure:

1. Know who is the maintenance person and how to get him to the site as soon as possible.

The first call should be directed to the HVAC engineer to describe the problem in as much detail as possible. This helps the engineer determine if it is necessary to bring the component. In the first time of the failure, it must be counted every second.

2. Understand and predict what happens if a refrigeration unit fails

For example, how long the server can last before the room temperature rises to a critical value (>120°F) You can calculate the maximum time left for the service person and whether you need to start the backup system.

3. Signing service level agreements with relevant companies on key environments

If you have not signed, it is best to prepare a portable refrigeration system. For example, Tripp Lite's SRCOOL12K is specifically designed for emergency cooling. It can provide approximately 12K BTU cooling capacity. In some places, emergency cooling equipment is very expensive and can be rented for a short period of time. If the room temperature reaches a critical value only a few hours before, you may face financial losses.

4, turn off non-essential servers

Development servers are usually used by large consumers, they do not need to run during production, test servers, and all non-essential servers should be closed Drop it.

5. If the temperature of the equipment room rises to a critical value, it is best to open the doors and windows. In short, the idea is to lower the temperature of the equipment room. If the outdoor temperature is lower than the indoor temperature, you can use a fan to blow out the hot air. .

6. The last one is the most important one to deal with the failure of the refrigeration system: ensure that all data is backed up in advance.

If the emergency plan for troubleshooting the refrigeration system is completed in advance, it will play a huge role when the real failure occurs. If you know that your server will not be able to shut down if the fault is not resolved, it is best to start the remote standby data center in advance and prepare for the transfer. The company's contingency plan should ensure that the business is functioning properly under all circumstances, even in very urgent situations. In addition, a disaster recovery plan should be developed to transfer services across the data center to the alternate data center in an emergency such as an emergency.

Server Refrigeration Redundancy

A refrigeration unit failure should not trap the entire data center into the oven. If the refrigeration system is redundant and has a corresponding failover plan, you can slow down Slowly wait for the maintenance engineer to come to the door.

The most basic redundant cooling method is n+1, but a large part depends on cooling needs and the company's budget. For a medium-sized data center (about 1000 square feet), an optional cooling solution There are a variety of things, for example, you can deploy a 1.5-ton AC device to the server room, plus a 1-ton device as a backup, these devices can be load balanced.

ESPO's network architect David Langlands said: "For medium-sized environments, it's best to find a certified HVAC expert to communicate. If funding allows, a complete n+1 redundant environment is given. There will be 3-4 refrigeration units in operation, which means your refrigeration system will remain 100% available."

For large data centers (greater than 5,000 square feet), cooling needs are different. Compared to small and medium-sized data centers, large data centers generally incorporate designs to prevent cooling failures. Only then, whether it is normal routine Maintenance or troubleshooting, even during peak business hours, even if multiple systems are running, there is sufficient maintenance time guarantee. For example, Emerson Electric or some well-known independent consultants can help you design a redundant server cooling solution.

Large data centers generally use the following refrigeration technologies:

- Raise the space under the floor to enhance the flow of cold air.

- Install a baffle in the gap between the servers to increase air flow efficiency.

- The air in the entire equipment room is all from the cold aisle. All hot air is removed from the hot aisle. For large computer rooms, multiple hot and cold passages should be installed.

- A cooler mounted on the roof pushes glycol through a closed pipe into the data center and circulates in the data center for cooling purposes.

- Use an AC blower to blow air cooled by glycol into the server room.

There are other types of refrigeration equipment and technology. It is best to listen to the opinions of HVAC experts who can help you choose the most suitable product.

Best Practices for Server Cooling

Finally, we have summarized several important principles for dealing with data center cooling issues:

- Add HVAC expert numbers to speed dial groups . There is a refrigeration redundancy ready plan.

- Monitor the temperature and establish a system to record the temperature change. When the server temperature exceeds the warning threshold, the AVTECH TemPageR is such a system. It displays the temperature value in log and graph. The $200 device is simply worth a million.

- In addition to temperature, monitor humidity and do not let the server room freeze.

- See where all the water is going, whether it's a drip tray or a water route, just pull the water out of the machine room and it won't leak.

- Data center demand for refrigeration is getting higher and higher, but the growth rate of cooling demand is often neglected. Langlands said: "Do not design refrigeration system according to current needs, and should make future cooling capacity increase in advance. planning".

Summary

There are too many factors to consider when planning or dealing with server cooling system failures. The best way is to prevent problems before they happen. In one sentence, the secret is: Make redundancy and backups, and keep in mind the contact details of HVAC experts.