2013 can your data center survive the earthquake?

According to a standard Federal Emergency Management Agency report, 38 U.S. states have moderate earthquake risk, 60% of the United States population living in the earthquake risk of moderate to high levels of the region. When it comes to building data centers in countries with multiple earthquakes, we must always remind ourselves of two things:

(1) Whether the location of the data center will be affected by the earthquake

(2) earthquakes can happen tomorrow.

Figure 1: Earthquake risk map for the United States

The quake will invalidate the data center without warning, destroying the data center's building and its equipment, and even destroying the entire line of the data center. The most likely cause of the quake's failure was the destruction of equipment and building elements, such as floors and ceilings, that were installed in the data center.

Equipment risk

Even in modern buildings that meet the current design and construction specifications, it is common for some devices to fail to function properly due to earthquakes. In most cases, making buildings more resilient can be fairly easy and relatively inexpensive to mitigate the constraints and effects of earthquakes.

Figure 2: Destroyed computer and communication rack

A more powerful social impact than the damage to the facility.

Even if buildings and equipment are in shock, it is common for other related equipment to fail to run. For example, many key facilities rely on HVAC systems, using evaporative cooling to make use of community water supplies. If the supply is interrupted, even if the device is not damaged, it can cause days or weeks of no effect. The same is true for business: large earthquakes can cause temporary blackouts in large areas and can lead to prolonged interruptions due to strong vibrations. In a major earthquake in Southern California, for example, most customers expect water outages for more than a week and the worst-affected areas to be disrupted for days or weeks.

Facility Backup can reduce the chance of disruption caused by earthquakes. However, if the backup is located near the main site of the datacenter, it may not work because of the same earthquake. The closer they are, the more likely they are to fail for the same reason: they are more likely to be shaken, or both lines may be damaged, or fires need to be evacuated. There is a clear trade-off between proximity to the facility backup and risk. Closer backup locations can reduce logistical backup costs, but this also means greater likelihood of total failure. Depending on the site location of the two data centers, the "proximity" may be hundreds of miles or more.

Figure 3: Analysis of a California earthquake intensity

Using risk analysis

Other causes of operation failure can be foreseen and remedied. We can even conduct engineering risk analysis. Experts can calculate the probability that each earthquake may affect the failure of a facility to run, and the specific length of time that a device cannot function properly due to an earthquake. The same calculation can be done by assuming certain modifications, such as anchoring equipment, increasing emergency generators or on-site water supply, and so on. By estimating and assuming related risks, it is possible to make informed decisions better and to help repair the costs and benefits involved. And make contingency plans based on these issues.

In California, a http://www.aliyun.com/zixun/aggregation/17543.html "> utility, its management commissioned research into the possibility of losing all of its data centers in an earthquake." They learned that most of their risk came from inadequate seismic equipment and floors. After a while, they repaired some of the most dangerous elements and worked out plans for the unresolved projects. The fix decision is based on:

(1) Criticality risk factors

(2) complexity and cost of repairs

(3) The estimated service life of a building or equipment, for example, that is scheduled to be scrapped next year may not be repaired and the risk of failure may be managed

In short, once you figure out which scheme is cheaper: a patch before an earthquake or a loss of function after a quake. By calculating its seismic risk, companies can determine which failures are most likely and which are the most cost-effective solutions. Then they can fix and develop contingency plans to address the remaining risks. The bottom line is that if you are doing business in a region of multiple earthquakes, it is important to have earthquake risk management to know your own earthquake risk.