Because of this the industry now considers a more pragmatic approach to data centre service reliability and availability: a data centre design which is fit-for-purpose, coupled with well-designed technology, and wrapped with operations and management processes to deliver the outcomes for the customer.
If you put aside what some may consider to be a religious war and instead look at where to spend your investment money for delivering services to your customers, consider this…
According to Gartner, unplanned outages only account for approximately 5% of all system outages – the generally accepted leading cause of which, depending on whose research you read, is human error, and ranges between 60% and 80%.
This is followed by hardware failures, software failure, security error and, finally, environmental causes in low single digits; yet customers have insisted on data centre providers investing significant sums to mitigate low single digit percentages of outages.
Wouldn’t a better investment of services spend be on building smarter systems to minimise both planned and unplanned outages, including designing resilient infrastructure which leverage 'location-less' concepts?
Given that one of the most common causes of data centre related outages is human error, which is something that cannot easily be accounted for in a data centre design, isn't the focus on a data centre tier rating misplaced?
There is no question that data centre design plays a role in the availability and reliability discussion, but in reality it is a small part of the overall reliability equation.
Operations, process, maintenance, and quality and risk management strategies are by far the bigger part. This would seem to be borne out by UTI themselves introducing the Management and Operations stamp of approval in 2010.
The 'location-less' model is prevalent in 'cloud' systems, including Google, eBay, or Netflix, which are built upon systems that are agnostic of location, and delivered from data centres which are not necessarily high up in the tier classifications; rather they use the technology to deliver the required availability instead of relying of the ‘container’ they reside in – Amazon refer to this approach as Availability Groups.
Today in New Zealand, many organisations are demanding services from data centres based on a UTI tier rating.
This is hindered by a lack of facilities formally being certified against the UTI Tier Certification of Design Documents (TCDD), or Tier Certification of Constructed Facility (TCCF) at any tier level, despite some claims to the contrary - it is further confused by claims of certification to non-existent UTI standards, including Tier 3+.
Unless planning on solely purchasing square metres, shouldn’t customers consider buying a service availability outcome for the service they are consuming, rather than being specific about how that outcome is delivered from within the data centre?
At the end of the day what is more important in both a 'cloud' and heritage system is implementing infrastructure to exploit the technologies available.
Through careful infrastructure design, mainframe systems are running in New Zealand today delivering on service availability and reliability of non-functional requirements indistinguishable from 'cloud'-based systems.
In summary, the age old adage of 'judging a book by its cover' is equally applicable to data centre services - the discussion really should not be about the 'four walls' which services are delivered from; it is more about the design and execution of what is on the inside - the people, the processes, and the technology.
These three elements contribute more every day to the delivery of reliable and highly available services than the four walls ever will.