November 4, 2013
To paraphrase Oscar Mayer, “Everybody doesn’t like something, but nobody likes RFP’s”. For those of you who have never had the pleasure of responding to one, think of them as a prospective customer’s “Christmas List” for their next data center. They are time consuming for the prospective customer to put together and the typical provider submission resembles the phone directory of a small city. Ah, good times, good times. Unfortunately, for both parties, they are a necessary evil and they are still the best way for a company to evaluate their list of suitors. In other words, if you have to do it, you might as well do it right. Therefore, it’s surprising to me that although they may document everything down to the color of the tile in the men’s room, rarely do they ask the responding data center provider to deliver proof of the claims that they make within the document. This has always struck me as a little surprising. I mean sure, I like a nice off-white subway tile in the bathroom but shouldn’t I be more interested in the fact that the company that I’m about to give a very large check to is actually telling me the truth about the data center I’m buying? So, for all of you who think that a men’s room is just a men’s room, I’ve put together a list of questions that should be included in every RFP to ensure that your provider is able to back up their claims. While this should be a relatively non-threatening addition to your next Request for Proposal, more than a few providers out there might term these, “The questions we pray they don’t ask”.
Question One: Is the facility Uptime Institute design certified?
Providers can claim that their design is any number of things, including that it exceeds standards to merit its on self-designated “plus” rating. If the proposed design hasn’t received its verifiable certification from UI, the customer has to determine if they are willing to “trust” the provider’s claims. If they say they are ANSI T14942 Tier III certified, that’s impossible as the tiers in ANSI are footnoted as not part of the standard.
Question Two: Will the facility be certified by UI as a constructed facility?
The inclusion of this question has one specific purpose—to ensure that the facility that the customer purchases is the one that is actually delivered. A lot can happen between design and delivery, and in case of Tier certifications it usually does. This is why there are only three (3) design and constructed data centers in the U.S.
Question Three: If the answers to questions 1 and 2 are no (very likely), will you warrant that you do not have a single point of failure in your facility?
This question is designed to provide assurance that if the provider is going to use a non-standard design that the customer understands that it does not include a single “choke point” that can bring their operations down. This is where most folks who “disagree” with the UI fall down. Have only one chilled water loop? Uh-oh.
Question Four: If there is a single point of failure, how long, and how often, are the shutdown windows for it when it needs to be maintained?
This is the “okay, just what exactly am I getting into?” question. “We don’t need to shut it down” is not exactly a best practice for a mission critical facility.
Question Five: Has all of the current facility been IST commissioned (including expansions)?
This question allows the prospective customer to determine if their new facility will be tested under load, and in all failure scenarios, to determine that each individual system works as it should, but also, that the all systems work together as they should prior to moving in. Without this, you’ll never know if it works until the utility has its first outage while the data center has loads of 60%+.
Question Six: If not, when was the last first commissioning, and what was the total load tested?
This is a very nice way of asking if the data center was ever commissioned.
Question Seven: Will you rep and warrant that no additional equipment has been, or will be, added without a full IST?
This is a critical question. In many single backplane configurations, or “phased builds” full Level 5 testing cannot be performed without bringing down all previously connected data centers.
Question Eight: If the answer to question 7 is yes, is the IST expected to include failure scenario testing?
If the answer to this question isn’t yes, this should be a cause for concern as to the resiliency of the site.
The RFP process is critical for customers to identify the data center provider who can best address their needs. The inclusion of the questions I’ve suggested enables them to use their RFP to not only evaluate providers on a purely technical basis but on the overall performance of the site as well. The mission critical nature of today’s data center operations requires that customers be able to make their decisions on verifiable information rather than potentially specious claims.