Datacenter Redundancy N+1, N+2 vs. 2N vs. 2N+1

While it might be one of the most annoying things on earth when it comes to a parent being asked “are we there yet” a dozen times within a 30-minute ride, redundancy is everything for data centers. Redundancy systems can prevent outages from having a negative impact on a data center’s reputation, reliability, business operations, and ultimately their financial bottom line. What does redundancy entail? Why’s it important? What are the redundancy system options for a data center? Let’s explore.

What Is Data Center Redundancy And Why’s It Necessary?

So, what is redundancy in technical engineering terms? It’s basically a failsafe or backup in the form of a system’s critical components/functions being duplicated in ways that bolster the system’s reliability. In other words, the system is more likely not to fail when the primary source of power goes out.

Now, how does redundancy apply to data centers? The focus here is on the amount of spare power needed to provide customers a backup power supply when power outages occur, which is a huge causative factor in datacenter downtime. No data center likes to hear the word “downtime,” but it’s a reality that happens to data centers across the U.S. every year.

A 2013 study on data center outages by the Ponemon Institute surveyed 584 entities with some operational responsibilities for data centers. The findings showed the following statistics:

• 85 percent experienced loss of primary utility power within the last two years.

• Unplanned outages were experienced by 91% of the above respondents.

• The average data center shutdown time averaged two hours.

• Of those two-hour shutdowns, the downtime averaged over 90 minutes during each failure.

Why do outages happen? Most will immediately say weather, right? But, weather isn’t the only source of outages. From internal or external equipment failure to someone hitting a power line with heavy outdoor equipment, the possibility for an unplanned outage means that failure can happen to any data center and at any time.

Outages can cost data centers significant revenue losses, particularly if they’re driven by internet sales that require continual connectivity. Per hour of downtime each year, the average data center loses $138,000 in revenue. For big business like Amazon, those lost revenue numbers soar to over $1,000 per downtime second.

Redundancy Matters For Data Centers

The Ponemon study goes a long way in speaking to the importance of data centers everywhere implementing DCIM (Data Center Infrastructure Management.) If revenue matters, then downtime matters. If downtime matters, then prevention matters. If prevention matters, then redundancy is a must.

Larger entities generally have their servers at Tier 3 or Tier 4 data centers who counter unforeseen power outages with “sufficient” redundancy power systems. That said, not all these redundancy systems are equal in terms of value and protection. Look to how the system is set up to determine the degree of failsafe protection it offers, specifically whether the redundancy system is a N+1, 2N, or a 2N +1 setup.

N+1, 2N, and 2N+1 Redundancy Systems: What’s The Difference?

N+1 System

Imagine that you’re having a party with 20 invited guests. You’d naturally need 20 plates, right? That’s your “n” value. But, what if someone brings an unexpected guest or someone shows up out of the blue? The +1 accounts for extra plate you may need.

In data center language, it’s called a parallel redundancy – the number of UPS modules you need for essential connected systems.. plus one, and it offers a UPS (uninterruptible power supply) that is available 24/7. The result is a decreased chance of downtime.

N+1 systems do have redundant equipment, but the system is still operating on a common feed/circuit on at least one common point. This fact leaves the system open to failure. A fully redundant system has completely separated feeds as a failsafe. So, the degree of protection here is better than no failsafe, but things could be better.

2N System

Again, you’re having a party with the 20 guests. What a 2N does is double the expected number. So, you’d have 20 extra plates for your party, not just one.

For data centers, a 2N redundancy system means double the equipment that’s needed for essential operations. Each run separately with no possible common points of failure. It’s a fully redundant system capable of offering an independent failsafe should an extended power outage happen and you need to keep things running. Think of it like having a spare car ready to roll if your primary vehicle gets a flat tire.

2N+1 System

This redundancy system is a combination of the above two. It doubles the amount of equipment needed and also has an extra piece for good measure. It’s the most comprehensive redundancy system, but most data centers find the 2N to be adequate and more financially friendly.

The Largest Data Centers in the USA

America’s Massive Data Centers

Data Centers are massive infrastructures that serve as a home to a network or a company’s most critical systems. These centers host an astounding amount of data and information, which are significant in ensuring smooth and optimal daily operations. They lodge computing resources as well as critical telecommunications which may include the following—servers, databases, storage systems, software, applications, and access networks.

Companies use these data centers to collecting, processing, storing, filing, and distributing large amounts of data, which cannot be accommodated in a regular office site. The top ten facilities listed herewith are the largest data centers by square foot for the United States. Although there are many impressive facilities strung across the country, they all fail in comparison to the size of the data center found hereunder.

  • Digital Realty-Lakeside

Topping the list is one of the oldest and largest data centers in the world. Built in multiple stages as early as 1917, this fortress takes up 24 million square feet of data center properties. Digital Reality manages 145 sites throughout the globe. Their biggest site is impressive at 1.1 million square feet located at the Lakeside Technology Lakeside Technology Center in Chicago, covering an entire city block within the southern loop of the Central Business District. This facility was designed by Howard Von Doren Shaw to b the corporate headquarter of printing giant RR Donnelley and Sons. It contained the printing presses that produced the popular Yellow Book (or pages) and the Sears Catalog. It is a Gothic type of architecture fortified by 4,675 steel-reinforced columns to carry the load of 10 to 12 inch floors that can also support 250 pound per square foot. This kind of structural strength was essential to accommodate massive amounts of paper weight stored in the upper levels. With 14 foot ceilings, this structure can take in heavy, bulky equipment, like transformers and various production machinery. The site has 21 vertical shafts for ease of transferring loads from level to level. Currently, these shafts house the fiber riser and power cabling. With its own fiber vaults numbering to 4 and three electrical power feeds, that allows it to provide the whole facility with 100 megawatts of power.

  • The NSA-Bumblehive

The first Intelligence Community Comprehensive National Cyber-security Initiative, with the acronym of IC CNCI, has its first center at the Bumblehive known as the NSA. This is the largest standing one with over one million square feet. This gargantuan building complex includes many functional sub units: from chiller plants, water treatment facilities, fire pump house, and its own electric substation. On top of that, there is a vehicle inspection facility, along with a visitor control center. Because such a massive center requires a lot of power to keep it running, there is even a sixty diesel-fueled generators on standby for emergencies that are capable of supplying the facility for three days with a hundred percent back up capacity. This primary purpose of this data center is to cope with the increase and influx of digital data, which is the result of the leaps and bounds in global networking. The goal of the NSA data center is to keep track of all forms of communication. Such a massive and ambitious undertaking which include tracking contents of private emails, internet searches, phone calls, travel itineraries, parking receipts, and the like! It feels as nothing is too small and no data is to minute not to notice.

  • QTS Metro-Atlanta, GA

Standing at 970,000 square feet, the third biggest facility to land on the list is the QTS Atlanta Metro. With a data center foot print of 530,000, it indeed belongs to the list of gigantic data center. Because of its size, it has its own fully operational, on-site Georgia Power substation. On top of that, they have direct fiber access to multiple technological and communication carriers.

  • IO- Edison, NJ

One of the largest modular date centers in the United States, IO in Edison, New Jersey takes up 830,000 square feet. Noteworthy, this is a former printing press facility of the famous daily, the New York Times. This is also one of the few data centers that were able to run almost immediately because of its modular technology. Perhaps, the largest advantage and perk of this data center that sets it apart from the other is prime location. Its location next to a large power switching station also poses an advantage. IO New Jersey boasts of fiber optic connectivity from no less than two of the largest IP backbones around the globe.

  • Terremark Worldwide-Miami

In operation since 2001, this large data center located in bustling Miami occupies 750,000 square feet of total data center footprint. Given its location, this six story building was fortified to withstand hurricanes. They have served as a connection hub for Latin America as well as the South Easter United States. Converging at this building are 160 networks, which makes a thriving connectivity ecosystem for its unique target audience. On top of the roof’s facility are three large globes, which conveniently store two 16-meter satellite dishes, and one 14-meter dish. The latter was set-up to provide backup connectivity for its esteemed clientele should there be an off chance that the facility loses its fiber needs.

  • Microsoft- Chicago, IL

Everyone knows Microsoft and it is not surprising for anyone to read this large software giant, who was one of the pioneers of the industry, belonging to this list. With around 700,000 square feet of space, this center is not the cookie cutter sort. Its unique design makes it atypical when compared with other bustling data centers. Built to accommodate 40 foot storage containers is what’s in the first floor. But they aren’t really meant to store containers. Instead, they are packed with web servers. The upper level contains the tradition space for all the data needs.

  • IOS- Pheonix, AZ pheoneix az 

Nixing the coveted 7th spot of the list is another IO facility, but this time it is located in Phoenix, Arizona. With a compound occupying 538,000 square feet of space, they have the capability to house both the data center and the main headquarters in this one location. The main highlight is actually not the size of the data center, but its rooftop solar paneling. They are one of the primary companies to utilize this technology. Their solar powered system has the ability to supply 4.5 megawatts of energy for the facility.

  • Apple-Maiden, NC

This sits on 2183 acres of land that Apple purchased specifically for this purpose. Move over Google and Microsoft! The buzz around town says that they are looking to add another 75 acres, specifically the lots sitting across the road. What makes this site remarkable is its capacity to accommodate drastic expansions. A large private solar panel system was set-up to power this data center.

  • Microsoft- Quincy, WA

This is the second Microsoft center in the top 10 list. With 470,000 square feet, the Microsoft Quincy data center is the home of the equipment and technology powering one of Microsoft’s newest babies, the new Windows Azure. This is a cloud development platform that’s innovative and functional. This center has a lot of storage capacity and can hold 3.7 trillion photos.

  • DuPont Fabros-Ashburn, VA

Last but most definitely not the least is DuPont Fabros in the Loudon Metro Data Center located in Ashburn, Virginia. Spanning 416,209 square feet is only a drop in the bucket for DuPont Fabros, who already owns and operates at least a half dozen data centers in the area.

How Many 1U Servers will Fit in a Rack?

Just like warehouses store products, data centers are warehouses for digital information. They contain things like servers that are used to host websites. Businesses often outsource their web hosting, data storage, and information technology needs to data centers because it’s cheaper, data centers are more reliable than most in-house operations, and they’re often more secure than businesses’ in-house operations.

One of the most prevalent components of data centers is servers used to host websites. These servers are frequently stacked on top of one another to save room on large server racks that are dedicated to the exclusive storage of servers.

To answer the question of how many 1U servers will fit in a server rack, we first need to define what 1U means.

What is a standard rack unit?

Just like temperature and pressure are measured with degrees of Celcius or Fahrenheit and pascals, respectively, standard rack units are used to indicate how large servers are.

In the field of data center colocation, the sizes of servers are expressed in standard rack units, which are often abbreviated as U or RU, respectively.

Virtually all modern servers are the same width and length. The only thing different about their size is how tall they are. One unit or rack unit equates to a height of 1.75 inches, or 44.45 millimeters.

A server that is 1.75 inches tall is given a rack unit size of 1U. One that is seven inches tall, for example, is given a rack unit size of 4U.

How big are server racks?

Just how servers come in different sizes, racks come in different sizes, as well. Racks, which are almost always constructed entirely out of metal, are sturdy frames that hold individual servers.

The three main sizes of server racks of full racks, half racks, and quarter racks.

Full racks hold 42 rack units’ worth of servers stacked on top of one another. In other words, full racks hold about six feet of servers in terms of height.

Half racks hold anywhere between 18 rack units and 22 rack units, or roughly three feet of servers stacked on top of one another.

Quarter racks generally hold between 10 and 12 rack units’ worth of servers, which comes out to about 1.5 feet of servers.

Server racks also feature side-by-side columns to hold servers with. When shopping for racks, data center managers can simply refer to racks’ specifications to figure out whether they have multiple columns or just one.

Why are there different sizes of racks to hold servers?

Some facilities have very high ceilings. In such facilities, data centers are able to store dozens, if not hundreds, of servers stacked on top of one another. Other facilities have standard eight-foot or 10-foot ceilings, in which managers usually don’t house anything taller than single full racks.

Data centers need to know what size various racks are in order to plan out their server-hosting infrastructure. The aforementioned units of measurement exist to make such planning easier.