Arc Flash: Fact Sheet for the Rising Data Center Threat

Details

Created: Wednesday, 07 December 2016 17:35

Special Thanks to Doug Christianson, EMT Corp., for many contributions to this fact sheet.

At one point in the not-too-distant past, 100kVA was considered a LOT of power for a data center. In today’s measurements, a 1MW UPS system with 300kVA PDUs supplied by 480-600V feeders is considered commonplace. Even in our homes, electrical service has grown from 60 amps to over 200 amps in just the past few years.

With growing power requirements comes growing disaster possibilities.

An arc flash incident has always been one of those possibilities, but especially recently with the power demand growth and the constant need for bigger and more sustainable equipment, arc flashes have grown more common and more deadly.

Some arc flash facts:

• 5-10 arc flash explosions occur daily
• Over 2000 people are treated annually for arc flash burns and damage
• The heat discharge from an arc flash can reach 35,000°F, almost 4x the temperature of the surface of the sun
• Generally, electrical systems need to operate at 480V or higher for an arc flash to occur
  
  

What is an arc flash?

In simple terms, an arc flash is an undesired electrical short circuit from one conductor to another or to a ground.

Electrical faults occur fairly often – a breaker or fuse may have tripped, the power can be interrupted. However, on occasion, prior to tripping, an arc flash event occurs.  Arc flashes specifically involve: melted metal, a high temperature, fast speed, and explosive discharge.

What causes an arc flash event?

The three main causes of almost any data center issue, including arc flash, are: Human error, inadequate PM, or improperly applied equipment. Arc flashes also can occur via the following:

• Accidental contact from worker
• Improperly applied or installed equipment
• Contamination across insulated surfaces
• Deterioration or corrosion of equipment
• Dust or condensation build up

Approximately two-thirds of all electrical injuries are due to the fault of the worker.

What happens during an arc flash?

An arc flash typically begins with a short circuit path between energized phases or phase to ground.  Since the load is removed from the path, extremely high fault currents cause the conductor metal to melt and convert to a gas.  During the phase change from solid to gas, the solid material (copper or aluminum) expands roughly 64,000 times original volume causing the arc flash explosion which consists of the following items:

The electric arc is enhanced by a massive blast of plasma and metal with extreme force and explosive damage.

The pressure from the arc blast can be detrimental enough to blow out eardrums and damage internal organs, as well as dispel shrapnel at extreme velocity.

The flash itself can be so bright as to cause temporary or even permanent blindness.

The blast can affect equipment all throughout the room, either by direct physical contact or through internal power damage downstream. In some cases, it can also be fatal to people in the room.

How to prevent an arc flash:

1. First and foremost, data center designs should be thoroughly studied with qualified facilities electricians.
   1. Which panels, switchgear, UPSs, ATSs, etc. need to be shut down for maintenance and repair?
   2. Which computing systems might be affected?
2. Complete an in-depth arc flash analysis.
   1. A comprehensive short-circuit study should be conducted for the data center space.
   2. A breaker coordination study must be done, even if one has been done before.
      1. Ensures that the breaker nearest to the fault trips before any unintended current can pass through.
3. Do not allow electricians to work on "live" systems.
   1. Power may have to be shut down for regular maintenance or observation, or even just to open the panel door.
4. Ensure that all personnel in contact with equipment are qualified and prepared to assess and handle different equipment with varying degrees of arc flash hazard levels.
   1. Proper gear and tools should be used at all times and regulations should be followed accurately.

How Redundant Can You Be?

Details

Created: Friday, 28 October 2016 16:24

Redundancy. It’s a term we hear less as a suggestion and more as a necessity for the data center space. Redundant data centers are less likely to have unexpected downtime, single points of failure, detrimental mechanical issues, or even problematic user error. But just how redundant should you be? Is there a golden standard? Instead of observing this question as how redundant you should be, ask yourself how redundant you can be.

Take a step back: What exactly is redundancy?

Imagine you are grilling burgers for a barbecue. You’ve totaled up your RSVPs and determined that 10 people would be attending. This is great! You know how many burgers to buy, how many to toss on the grill, and how many people will eat a burger. If everything goes according to plan, you will be able to spend the exact amount of time, money, and energy on 10 burgers with no portions lost.

If everything goes according to plan, of course.

Two very simple issues can happen here:

1. One of the burgers slips off the plate and plunges to the grass and dirt under the grill. You now only have 9 burgers but need to feed 10 people.
2. Someone who initially thought they couldn’t make it unexpectedly shows up due to a last minute schedule change. You now only have 10 burgers, but need to feed 11 people.

In the simplest of definitions, redundancy is preparing ahead for these issues by having more burgers than you need. In data center terms, this is like having more UPS units or more cooling units than you technically need in case of unforeseen issues.

Redundancy comes at a price, however. The more units you have to cover these potential problems equals more money to purchase, more energy used to maintain, higher output costs and measurements in electricity, etc.

N, N+1, and 2N: What do they mean?

To some, defining these calculations might seem novice, however it can be very easy to misunderstand what these numbers mean – sometimes we become so accustomed to a certain method that we forget to explore other options.

Let’s take a look at N. N represents the exact number of units that is sufficient to accomplish your goal, such as our 10 burgers for 10 people or the number of UPSs to cover the data center load.

Is N enough to sustain your data center? Technically speaking, yes. But without any redundancy, you are taking a huge risk and betting all your money on one horse. If anything happens with N, you are looking at downtime, extensive money loss, and incalculable stress.

N+1 was devised to protect against these issues – having one more unit than required in case of issues. If you had 11 burgers, you would easily be able to feed 10 people if one burger fell on the ground, or feed 11 people for the unexpected guest. Likewise, having one more UPS or cooling unit in place will protect your data center in case one of the other units goes down.

This isn’t just limited to +1 either. Some organizations use more units to cover more issues, such as N+2, +3, +4, etc. depending on the size and requirements of the room. The more burgers you have, the more able you will be to resolve issues of dropped patties or additional guests.

The “N-plus” factor is very useful if you feel extremely confident that you can predict these potential issues. But sometimes, problems occur outside of our control. What happens if you have 4 extra guests to you BBQ, but accidentally drop an entire plate of 5 burgers? Even with an N+4 system, you are still unable to feed 5 guests.

2N is the calculation that strives to cover all bases. Rather than adding individual units in case of issues, instead you double the required number on a completely separate power stream to ensure that you can bounce back from even a total loss.

If 10 people RSVP to your barbecue, this means you are grilling 20 burgers on 2 separate grills. Now you can drop a plate AND have unexpected guests and easily feed everyone. You can even accidentally knock over an entire grill’s worth of burgers while tossing the football around, and your other grill will run unaffected and pick up the slack. With a 2N set up, if N goes down entirely, the redundant N will remain unaffected.

In comparison, an N+1 system is usually running from the same power source and so if the entire system goes down, even the redundant parts might go with it.

And even still, if you are extremely worried about downtime and have the funding and manpower to maintain it, some organizations even go to a 2N+1 system – the ultimate in uptime stability. It goes without saying that this system, however, is absolutely a costly one and is most effective with industries such as banking, power, and government where uptime availability can be the difference between a thriving business and complete professional destruction.

Comparing and Contrasting Redundancy

Why would someone choose an N-only system?

It’s difficult to generalize into one or two reasons why all N-only organizations have chosen to remain that way. Perhaps it is a cost issue; it is certainly the most inexpensive route in the short-term to go N-only. Sometimes that is coupled with the thought that said organization is not profit-dependent on 100% uptime and availability. A smaller shop with only a server closet, no online store, and one POS system might not see N+1 as a logical option.

In a broad sense, the more that uptime/availability influences your company profit and brand loyalty, the more viable an N+1 or 2N system will become to you. But there are so many factors that also affect this decision, and so it is important to discuss your options with a consultant who has extensive experience in the field like Facility IT Group. Let us know if we can help.

PUE is O-U-T

Details

Created: Thursday, 29 September 2016 15:16

In case you were not aware, ASHRAE (the American Society of Heating, Refrigerating, and Air-Conditioning Engineers) just released their new ASHRAE 90.4 Energy Standard for Data Centers publication. In this publication, many changes have been made in regards to the standards to which data centers have been upheld over the past ten years, including the removal of Power Usage Effectiveness (PUE) from their standard calculations.

ASHRAE has been a global leader in determining and explaining the operating standards for data centers and other high-powered facilities for many decades. In 1959, the American Society of Heating and Air-Conditioning Engineers merged with the American Society of Refrigerating Engineers and ASHRAE was born, extending into three generations of designing proper data center standards upon which operators are generally required to follow.

It should be noted that ASHRAE’s publications are not actually requirements but rather suggestions. However, it is regular practice for building officials to reference and adhere to ASHRAE standards when determining code evaluations and permitting, as well as inspections.

So what is different with this 90.4 publication? Let’s take a look at a few key changes.

Standards are no longer considered “prescriptive-based”, but instead more “performance-based”.

For the past ten years, it has been accepted that data centers were required to meet certain calculations of energy efficiency and wasted power. Due to this, equipment designers and data center providers have been dedicated to reducing their waste and delivering the highest quality of efficiencies available.

This progress meant it was no longer about trying to achieve the standard, but instead about trying to achieve perfection. And with some of the “greenest” equipment seen in our time, ASHRAE decided that it is not quite as important to standardize the equipment as much as it is to standardize efficiency of the equipment.

For example, economizers have been pretty much required in order to meet the energy standards set in the 90.1 publication. Even if the economizers were unreliable or impractical to use, many felt obligated to install them to remain within the standard. Economizers have been especially troublesome for data centers in warmer climates where they didn’t see the return as well as colder-located facilities.

The new 90.4 standards allow data center operators to use other means of ensuring efficiency without mandating the equipment they use, so long as they meet the output calculation standard.

Upgrade projects now respect individual components rather than forcing unrealistic updates.

With a single-calculation system, this change can be difficult to understand. ASHRAE realized that in order to follow 90.1 standards with a major upgrade to a data center, many times it forced operators to upgrade other major components when it might not have been a realistic business decision to do so. But in order to comply with 90.1, data centers were stuck between a rock and a hard place.

The new 90.4 allows data centers to upgrade a component of the data center while allowing a tradeoff between mechanical and electrical infrastructure. For example: If you need to replace your UPS system, but doing so will mean that your cooling system will no longer fall under the cooling energy standards, you won’t be forced to upgrade your entire cooling system.

Instead you can purchase a highly efficient UPS and smaller components to help distribute power more efficiently (such as transformers). As long as the new electrical equipment works together with the existing mechanical cooling equipment to fall under the more general performance efficiency standards of 90.4, you will not be forced to go through the lengthy and unnecessary process of upgrading the cooling.

PUE has been removed as a standard calculation and is replaced with MLC and ELC calculations.

Power Usage Effectiveness standards have been wonderful for the data center community. It has forced operators to think logically and intentionally about the energy efficiency of their equipment and the amount of power they waste daily. Without PUE’s implementation in today’s data center world compared to when it was first introduced in 2006, the amount of wasted power and wasted money spent on energy would be astronomical.

However, in recent years with the amount of change in data center equipment and the constant need for upgrades, it has not made as much sense to use PUE as a standalone standard. PUE requires actual in-use operating data, which is very difficult to measure accurately. Issues with the power delivery company can cause PUE numbers to rise, even if the equipment is not losing much wasted power.

Also, server utilization plays an important role in how PUE is measured. For example, a company with high PUE and heavy server usage might actually be more efficient than one with low PUE and low usage. Plus, PUE is easily abused in favor of marketing, which can also make it very difficult to judge accurately.

To solve these problems, ASHRAE has removed PUE in favor of Mechanical Load Component (MLC) and Electrical Loss Component (ELC) calculations.

            Mechanical Load Component (MLC): This is a formula used to determine the efficiency factor which represents the amount of mechanical power needed for design data center load. This calculation standard also varies depending on the climate, making it easier to address the concerns of different physical locations.

            Electrical Loss Component (ELC): This is a calculation to determine the losses experienced through different segments of the electrical system. Each segment of the system is upheld to different loss characteristics (such as a single-phase vs. a 3-phase UPS), adjusting for varied sizes and configurations.

Rather than try to force these two calculations together, ASHRAE keeps them separate and holds them to their own qualifications. In doing so, it prevents businesses from being obligated to make certainly costly and unnecessary upgrades to keep up with PUE standards. This allows these facilities to still strive for the most energy efficient designs as possible, while also being realistic in terms of what to upgrade and what to retrofit.

This will be particularly helpful for legacy data centers where keeping up with PUE requirements has been detrimental to business; elaborate upgrades and design alterations to account for new technology has forced extremely costly solutions to maintain efficiency levels. The separation of MLC and ELC will allow these data centers to make the upgrades they need to as necessary and convenient.

Tradeoffs between MLC and ELC allow for a more lenient 80/20 policy.

Perhaps the greatest change of 90.4 is that they allow tradeoffs between MLC and ELC calculations. On their own, each MLC and ELC are upheld to a standard calculation that data centers must either equal or be less than in order to qualify. In contrast to PUE where operators pushed to have a lower number close to 1.0 without any real sense of exactly how efficient their system is, MLC and ELC provide the guidelines to more accurately gauge the efficiency levels of the data center.

Since each calculation is upheld to their own standard, the total efficiency target is merely a sum between the two numbers. ASHRAE determined that as long as a data center’s total calculation is equivalent to or less than this sum, they adhere to the standards set forth in 90.4.

What does this mean for a tradeoff? Imagine that your MLC calculation is below the standard, but your ELC calculation is above it – when you add the two together, the total is equal to the total standard. Even though your ELC is above the standard, the total still qualifies for the standard and therefore, no changes need to be made to meet 90.4 guidelines.

Previously, the above situation would likely mean a data center operator would be impelled to make changes to the electrical system, which may in turn require an unnecessary (if even small) change to the mechanical system to ensure compatibility. Now the operator can rest easy for a while and make the changes as seen fit.

These tradeoffs ensure that only the most out-of-date or inefficient systems will be forced to upgrade – seen as an 80/20 policy. 80% of current data centers will most likely fall under or equal to the standard because of the allowance of tradeoffs between MLC and ELC, and only about 20% will need to make upgrades to meet the criteria.

ASHRAE 90.4 has gone under several public revisions and has produced many addendums, as these standardizing publications regularly do. Currently the group The Green Grid (who devised PUE in 2006) is reviewing ASHRAE 90.4 and determining if and how to implement MLC and ELC. The removal of PUE will definitely not be an overnight change considering how prevalent the calculation has been over the past decade.

But it is important to understand the difference between PUE and MLC/ELC as we will likely be seeing much more of these calculations in the very short future.

You can purchase ASHRAE 90.4 here

Don't Be Afraid of the Cloud

Details

Created: Friday, 01 July 2016 15:16

One of the greatest factors preventing companies from moving any of their critical processes and data to the cloud is perhaps one of the simplest concepts: FEAR. And understandably so – it is good to be fearful of anything in control of your critical data, especially if it is something surrounded by much scrutiny.

The cloud is continuously heralded as the future of IT with more and more big names jumping on board and ongoing creation of new associated programs and technology. But that still doesn’t answer the biggest fearful question: “How do I know my data is safe?”

Despite the constant scrutiny, cloud security is an incredible asset with extreme power.

Provides Many of the Same Functionalities as Traditional IT Security

• Protects your information from theft
• Prevents any data leakage
• Shields against unwanted deletion
• Servers are stored and protected within cloud data centers

Compliance Requirements in Place to Protect Critical Information

• Cloud data centers likely have stricter regulatory standards than traditional IT security measures due to multi-tenant retention and the need to overcome regular inspection
• Support for compliance regulations such as PCI DSS, FISMA, FedRAMP, HIPAA, etc.
• Extra security programs go above and beyond to keep your information safe
• With clients such as financial institutions, medical centers, and other enterprises with valuable personal information, customer confidentiality is an utmost priority
• Since cloud data centers are focused completely on the performance and security of their infrastructure, they are better prepared against attacks than traditional in-house IT

Of course, people do not often enjoy losing “control” of their property. One criticism against cloud is that giving up the control of physical equipment means a higher vulnerability. But it’s important to remember this:

Control does not mean security

Anything that can be accessed from the outside, enterprise OR cloud, has equal chances of being attacked. Attacks are generally attempted via favorable circumstances – if there is an opening in the wall, someone will try to get through it. Therefore, the physical location of your data matters much less than the means of access.

• Cloud providers go way above and beyond to secure client data, more than any client would typically do in-house
• Cloud security is not a product or a service, it is a process that is ongoing and always evolving

Leo Reiter, CTO of Nimbix, spoke on the topic of network security: “Cloud providers live, eat, and breathe network security while most other organizations don’t usually list it as one of their core competencies.” It is important to understand at just what length these providers are going to protect your data, and that even with an in-house data center, you are not completely protected in comparison to a cloud data center.

When it comes to the security of your mission critical data, it is more important than ever to educate your employees about what is expected in terms of confidentiality. Insider/employee attacks are the third biggest threat to your data’s security. Keeping your employees aware of the security measures in place and ensuring their compliance will drastically reduce the level of threat to your processes.

Enterprises like Netflix use cloud data centers, despite having hundreds of thousands of customers’ personal data on file. As of January 2016, they have shut down their last data center and are completely running off public cloud. You can read more about their transition here.

Evolve IP is one of those companies that lives and breathes cloud. They have partnered with top security companies such as Cisco, RSA, and Qualsys to provide powerful security solutions. Some of their network security options include:

• Security Monitoring and Management
  • 24x7 monitoring, event notification, management, and fault resolution
• Security Event Management (SEM) Service
  • Extends firewall and IDS capabilities – analyzes data in real time and automates the process
• Cloud Firewall
  • Network-based private firewall – eliminates traffic to premise, increases bandwidth
• Cloud IDS
  • Network-based active intrusion detection
• Managed Firewall
  • Premise-based firewall, supports onsite DMZ or management requirements, monitored and managed from cloud
• Managed IDS
  • Premise-based active intrusion detection
• Threat/Vulnerability Scan
  • Identify vulnerabilities, ensure regulatory compliance, prioritize remediation – no infrastructure to deploy or manage
• PCI Compliance
  • Automated PCI scanning from cloud, backed by certified auditing and consulting services

And this is only the tip of the iceberg. There is a multitude of different security solutions and options available to you. All it takes to get started is some research and the willingness to look past the scrutiny to see what works for yourself.

It’s time to start looking into how bringing some of your processes to the cloud can benefit your business. We can help you take a deeper look into which provider to use and the protection you will have against attacks. Give us a call and let’s get started!

Evolve IP is The Cloud Services Company™. Designed from the beginning to provide organizations with a uni¬fied option for cloud services, Evolve IP enables decision-makers to migrate all or select IT technologies to its award-winning cloud platform. Evolve IP’s combination of security, stability, scalability and lower total cost of ownership is fundamentally superior to outdated legacy systems and other cloud offerings. Today, over 80,000 users across the globe depend daily on Evolve IP for cloud services like virtual servers, desktop services, disaster recovery, unified communications, contact centers and more.

Evolve IP’s Network Security solutions offer businesses a better way to protect their critical business data by delivering turnkey network protection from both internal and external threats without the expense or complication of on-site devices or software.

Sources:

https://cloudsecurityalliance.org/
http://www.cio.com/article/2922374/cloud-security/20-of-the-greatest-myths-of-cloud-security.html
http://www.techrepublic.com/article/cloud-security-10-things-you-need-to-know/
https://aws.amazon.com/security/introduction-to-cloud-security/
http://searchcloudcomputing.techtarget.com/opinion/Clouds-are-more-secure-than-traditional-IT-systems-and-heres-why