So, we now know that "Tier" ratings come in two versions: TIA-942-2 (TIA-942) and Uptime Institute Tier Standard: Topology (TS:T).

But are they really different?  And if so, what are the core differences.


Well, they are different, and in some subtle but very important ways.

First, Both examine the resiliency of Power and Cooling systems.  However, TIA-942 also includes the categories of Architectural and Telecommunications.  (That should be no surprise, as they are a Telecommunications Standards body).  In Truth, TS:T also includes telecommunications (cabling) to a limited degree, where multiple redundant routes are concerned (at Tier III) and routing compartmentalization (at Tier IV).  But that is the extent to which TS:T consider telecommunications.

The real core difference between the two, at the theoretical level, is what is considered the Primary source of Power to the site.  TIA-942 treats Utility Power Supply as the primary source for power.  At Tier 3 it requires N+1 Utility feed (so, a minimum of 2), and at Tier 4, 2N (Also, minimum of 2, but could require more, if 1 feed each will not suffice to provide total power to the site).  Generator power is used as a "bridge" only in the event that all Utility power is lost, with the expectation that utility power will be restored in a time shorter than there is fuel supply available to the generators.

TS:T treats on-site power generation as primary power, with utility power used as a low-cost alternative when available.  For Tier II and above, that requires redundancy in the generator set, but all are able to utilize N+1 generator set for redundancy, so long as the proper rules for down-stream power distribution are met which also provides for multiple power supply pathways under Tier III and Tier IV.  This carries some distinctions.  Under TS:T then, it is technically possible to achieve any Tier level with 0 incoming utility power feeds (So long as the site is able to provide continuous power generation, with redundancy for Tier II and above).  One key "gotcha" then when building a TS:T standard data center is to ensure that the generators are rated for either PRIME or CONTINUOUS run (as opposed to STANDBY rated).  If a PRIME generator is used, it must be de-rated by 70% of its capacity to meet the continuous run requirements (however, if agreed in writing with the manufacturer, it is possible to achieve up to 85% capacity for de-rating).  STANDBY generators are not robust enough for Tier III or Tier IV, but are suitable for Tier I and Tier II.  Generally, STANDBY generators are rated only to run 200 to 500 hours per year.  (200 hours is 8 1/3 days, so by the time you reach day 9, you've hit maximum for the year.)


This distinction may seem trivial, however it is a crucial difference in the underpinning philosophy of TIA-942 compared to TS:T.

Best Regards,
Scott

"Everything should be made as simple as possible, and no simpler."

Dear Scott ,

would you please provide more clarity around the engine generator plant and whether to use single ATS for the load inside the Data Center (UPS+Cooling system , etc..) or its better to segregate the load e.g one ATS for UPS , another one for Cooling system>

Thanks

Ksomer

Ksomer,
   Can you tell me more about what your Tier objective, or resiliency objective is?  The answer to this question will depend greatly on what it is you are trying to achieve.
 

Best Regards,
Scott
ATS, CDCE

"Everything should be made as simple as possible, and no simpler."

I am beginning to believe that Tier's are great for very large or hosting organizations but maybe a different model would work better for the small to mid-sized company that wants to own their own data center.

Is it more effective to have one or more Tier 3/4 very expensive data centers or many smaller less expensive Tier 1/2 facilities in a metro area configuration? The latter would require a more refined application resiliency but I believe it would be more cost effective and arguably increase overall fault tolerance.   

http://www.datacenterinfoblog.com/

DCIB,

Quote:

Is it more effective to have one or more Tier 3/4 very expensive data centers or many smaller less expensive Tier 1/2 facilities in a metro area configuration? The latter would require a more refined application resiliency but I believe it would be more cost effective and arguably increase overall fault tolerance.

   You ask a great question.  First, its never a good plan to have only 1 data center.  Tier III and Tier IV are about high availability, they are not, and do not guarantee business continuity.  The choice to have any Tier III or IV DC should be based on business drivers, and the need for high availability.  This is something I see often, and is a poor strategy: "We need a Tier III DC, because we have high availability applications.  So let's build one to hold everything".
   This is a flawed strategy.  Many organizations that build Tier III or Tier IV DC do have a need for this high availability, but only put around 15% - 20% of Tier III applications in it.  The rest of the space is used by Tier II or Tier I equipment.  This is a big waste of money, and then when it gets full they go, "Oh, we need another Tier III data center".  Reality is, a second DC, in Tier II or Tier I is much cheaper, and they can move the 80% of non-critical applications out of the Tier III, and discover instant Tier III space.
    There is some discussion about are two Tier I's, data replicated, more resilience than one Tier IV?  The answer, at the critical systems is, yet.  But you now have double the IT gear costs for that resilience, and the added complexity of the data replication, and cost of data transfer, and overcoming latency.  But cost would be about the same actually.  It's just not yet a common practice, and could require more power usage since now there is double IT load.
    One option is to colocate the Tier III and IV systems since most companies will actually need only a few racks worth (less than 10) to meet those critical applications.  Then, build a Tier I or Tier II site at lower cost, and (both operating and capital costs).  This could be a better strategy for most organizations that have small high availability needs, and large daily support (email, file storage, data warehousing, etc).

Best Regards,
Scott
ATS, CDCE

"Everything should be made as simple as possible, and no simpler."

Scott,

I just ran across your forum tonight as I was seeking info on Data Center tier designations. Thank you for sharing so much excellent info with us.  I have been an electrician for 32 years and now work (for 6 weeks) at a substantial DC in Phoenix.  We have one Tier 4 building and 1 Tier 1 building.  We also have 13 1640 kW generators available (I learned lots from your discussion on derating).  I have been around DCs for 20 years but I am looking to learn all I can for my new job which entails responsibilities for Electrical, Mechanical,and fire protection systems, among others.

Thanks again;  I'll keep on reading all that you have posted.  

Hi Scott
Thanks for your reply, actaully the objective is Teir II
Regards
Ksomer