Along with the increase of introduction of the cloud computing, various types of data centers are requested. For example, one with higher reliability, or one with more eco-friendly and low cost rather than reliability.
Among the standards that define the facility contents to realize required reliability when constructing a data center, “Tier” formulated by a private sector in U.S.A. (Uptime Institute) is known best. However, it is a facility standard built by reflecting global requirement and extensive consideration has not been given to the actual situation in Japan.
For example, as a basic concept on power infrastructure Tier considers in-house generation to be Primary and positioned commercial power only as its backup. Commercial power supply in Japan has a world-top level of reliability and we believe it is reasonable to take commercial power as the main source and in-house generation as its backup.
Another issue is that we need to consider the higher quality (low failure rate) of Japanese products and the need for earthquakeresistant specifications.
To resolve these general issues, JDCC has formulated the “Data Center Facility Standard,” with the aim of constructing a proprietary facility standard that conforms to the actual situation in Japan. This modifies those portions of the Tier standard that seem excessive for Japan and adds Japan-specific elements.
We plan to actively promote an understanding of this standard. Our goal is that an international understanding of this standard will help to enhance the international competitiveness of domestic data centers, and to have it widely adopted by parties related to data centers including data center users, designers of buildings and equipment, service personnel, construction contractors, and data center operators.
Reliability Comparison of the Commercial Power
Commercial power annual blackout (unit: min) |
Number of blackout time (time/year) |
End user annual blackout time per year (min/year) |
End user operation reliability | |
Japan(TEPCO) worst value in recent 10 years | 18 | 0.18 | 43 | 99.99% |
U.S.A | 86 | 0.86 | 206 | 99.96% |
England | 68 | 0.68 | 163 | 99.97% |
NOTE:
End user annual blackout time per year (method of calculation)
Calculated based on the assumption that one blackout causes 4 hours of fault for end users.
REFERENCE: Operational Reliability Listed in “Tier Classifications Define Site Infrastructure Performance (Uptime Institute)”
Total blackout hour of a enduser due to site (min/year) | Operation reliability of a enduser due to site | |
Tier 1 | 1728 | 99.67% |
Tier 2 | 1320 | 99.75% |
Tier 3 | 96 | 99.98% |
Tier 4 | 48 | 99.99% |
Conceptual Comparison of Uptime Tier (2008) and JDCC FS
Service Level of a Data Center Assumed at Each Tier Level
Service Levels | |
Tier 1 |
|
Tier 2 |
|
Tier 3 |
|
Tier 4 |
|
This standard is built by considering compatibility with existing facility standards such as FISC Standard*1 and JEITA Standards*2 that are influential in Japan, and taking overseas standards such as Uptime Tier*3 and TIA-942*4 , and provisions in ASHRAE*5 or IEEE*6 as guidelines.
Also, items such as floor height and power capacity that are irrelevant were classified as recommended items. Three items in the right are added as Japan original.
Though it is not reliability related, "Consideration for the environment” that is a social requirement for Data Centers is added as one of the recommended items.
Overall assessment of earthquake risk such as seismic hazard, stability of the ground, and earthquake resistance of the equipments concerning data center site.
If the building is dedicated for data center use. Or evaluation of the security levels, communication network, and facility levels.
Evaluation of the data center management system and operation management.
This standard defines reliability security required for data center facilities by separating “basic items” that are minimum requirements and “recommended items” that are expected to be employed for reliability security.
Upon the application of this standard, “basic items” need to be satisfied for all evaluation items specified in each level of “Tier 1” to “Tier 4”. Concerning the recommended items, not all the items have to be satisfied. Each DC can choose any items thought to be required according to the required reliability.
Along with growing introduction of cloud computing, various types of DC are requested such as one seeking the higher reliability, or one pursuing the cost performance or greening rather than the reliability. To respond to these requirements, this standard allows “Multi-tier data center” where each server room in the same data center can have different tier level as well as treating a whole data center as one tier level.
It is thought that Japan is a country of earthquakes and has higher earthquake risk. However, if we calculate the earthquake strength (maximum acceleration) as a seismic hazard in California that occurs once per 475 years, we get “480cm/sec2” and it is the similar strength(earthquake risk) as in Japan.
On the other hand, Building Standard Law in Japan is seemed as the most stringent building standard in the world. If we calculate PML* for the same buildings built according to laws in U.S.A and Japan respectively, Japan gives lower PML than California, i.e. lower earthquake risk. This means that data center built in Japan has higher security against earthquake than in U.S.A.(California).
Evaluation Result of Earthquake Risk in California and Japan (PML Comparison)
PML Estimation Result (RC Bldg, 8F) | |
California | 19.8% |
Japan | About 15% |
* PML is an acronym of “Probable Maximum Loss" which indicates the damage of an earthquake by a ratio of cost to recover from the damage (including compensation of non-use) vs. the original asset value.
* Lower PML means lower earthquake risk.
Reliability of major equipments used in a data centers such as UPS (Uninterruptible Power Supply) and air conditioner, is a significant factor that affects the reliability of the data center facility.
Data centers in Japan generally employ Japan made UPS air conditioners in view of their lower failure rate or higher efficiency. High reliability of these products leads to high reliability of a data center as a whole.
UPS usually used in overseas is said to fail by some reason about once per year at least. Time required to repair and recover is about 24 hours. These can be converted to operation reliability(ratio of continuous operation) of single UPS as 99.7260%(1 – 24/8760). UPS generally used in Japan has higher reliability with down time below 8 hours and gives 99.9087% (1 – 8/8760) operation reliability.
By this, in case of 4 + 1 unit configuration in Japan, operation reliability in Japan is 99.9992% which is the same reliability of 1 + 1 unit (2N) configuration in overseas.
Also, generally used UPS in Japan keeps 95% efficiency under the light load of about 30%, which is about 5% higher than the efficiency of those generally used in overseas.
Operation Reliability Comparison of UPS Systems
UPS System Configuration | Operation Reliability | |
Overseas (UPS generally used in overseas) UPS fault per year: 24H/unit | Japan (UPS generally used in Japan) UPS fault per year: 8H/unit | |
In case of 1 unit | 99.7260% | 99.9087% |
In case of 4+1 unit | 99.9925% | 99.9992% |
In case of 1 + 1 unit (2N) | 99.9992% | 99.9999% |
Comparison of UPS Efficiency
© 2016 JDCC All Rights Reserved