Archived Content

The following content is from an older version of this website, and may not display correctly.

Matthew Mescall, consultant, Uptime Institute Professional Services: The Uptime Institute released the latest version of the Tier Standard: Topology effective 1 August 2012. The updates to the Standard – last changed in 2010 – are the result of discussions and voting by the Owners Advisory Committee. The changes to the Standard reflect the experience of those organizations completing certification and the experience of Uptime Institute Professional Services which conducts the Tier Program for the Institute. Mostly administrative in nature, the changes to the Standard clarify the intent of the Institute regarding engine generators, communications, fuel oil systems, and makeup water for evaporative cooling systems.

Andrew Dewing, data center professional, UK: They are still advocating that data centers need to have generators as the prime power source and [be] rated to run continuously. Why do they want to design to the lowest common factor? For the vast majority of the planet with developed electrical utilities this is simply unjustifiable on every level. If you are designing a facility for a location with a utility of lower availability, then beef up the backup system design accordingly.

It sounds like UTI is framing its standard on experience in California, which is hardly an international approach.

Rhys Jones, senior electrical engineer, JCA Engineering: This generator thing seems negotiable – I saw the data for one site recently certified Tier III and the PRP load factor of the gen sets was either missed or ignored (or the vendors gave a written qualification that was more risk based than evidence based). The worrying thing here is that in each of the potential cases mentioned above, to my mind, there is an undermining effect of the process.

As you allude to, the assessors should understand the grid availability of the nation in which the site they are certifying is based.

I am also somewhat dismayed that there has been no real study completed – at least to my knowledge – whereby utility availability (by nation, region and so on) has been collated to give an indication of how firm the grid is (in terms of 0V events and sags/swells). Clearly, the risk/requirement for standby power for a data center in Zimbabwe is going to be somewhat different to that of the Netherlands.

Rowan Peck, director, Norman Disney & Young: When I did the Uptime ATD course last year, this point was discussed at some length. My understanding from that discussion, and from the (unchanged 2012) written word in the Tier Standard: Topology, is that UI do not require that a site runs on generators all the time (although for some remote sites, that is necessary).

The thinking goes like this:

  1. The EG system is the primary power source, and the grid is an economic alternative. In other words, the EG system has to be able to support the site, but most operators will of course use the grid if it is there.
  2. Hence the statement in UI that: “Disruptions to the utility power are not considered a failure, but rather an expected operational condition for which the site must be prepared.”
  3. Given that the EG system has to be able to support the site in the absence of grid, it had better meet the Tier requirements as well – which if you want Tier III, for example, means that limiting the run hours or not being concurrently maintainable are bad ideas. So don’t meet Tier III (or IV). For Tier IV, Fault Tolerant and Compartmentalization requirements for the EG system must also be met.

If I am designing or reviewing a site for Tier III or Tier IV, it’s not a case of how reliable the grid is, nor does a reliable grid mean something can be ‘left out’. That question is relevant to whether it’s economic to build in a given location to start with, taking whatever else into account. Zimbabwe may have poor grid reliability (I don’t know if it does or doesn’t) but still be the most economic location in the region for other reasons, such as fiber distance, politics, build costs and so on.

My read of the Tier Standard: Topology assessment question is whether the site meets the Tier Topology criteria or not – not how it will be powered under normal circumstances once it gets going.

Frank Bodi, reliability manager, Silcar, Australia: Rowan, it seems that terms like “concurrently maintainable” are being used as a substitute for concepts such as “reliability” and “statistical significance”. I have no problem with seeing the grid (in Australian CBD areas) as a concurrent element to a generator. Whether something is “concurrently maintainable” (generators or anything else) has only an indirect linkage to system reliability. Andrew and Rhys were alluding to this, I believe.

I think that concepts that started out as being generically good for the industry are possibly being taken out of context. From discussions I have had with those in the UI circle, the thing that appears to be absent is sound statistical information and application thereof – whether intended or not, it comes across as a one-size-fits-all approach.

Rowan Peck: I agree with most of what is being said but not if you’re asking me whether a site meets with UI tiers, or it does not. In that case, the UI standard is applied. There could also come comment on the side when commenting about UIness, about some of the situations you have put forward, Rhys. Ultimately UI certify sites or they do not.

We’re questioning the validity of the UI standard here, which I have no problem with. But right now, it’s the standard that many sites and operators chose to adopt. Time and the market will determine whether it needs to be amended in future to take into account grid availability (or lack thereof). Right now, grid availability is not a factor considered. There are plenty of good reasons why ignoring the grid does not make sense in some cases. Some customers in those situations decide UI is not for them. Others look at TIA, which has a slightly different take on the issue, and decide that TIA is for them. It’s not always a technical choice; sometimes it’s a branding one, for example.

This article first appeared in FOCUS 24. To read our digital editions, click here.