Posted by Kim Otte on Tue, Aug 03, 2010 @ 11:28 AM
The battery is the most vulnerable part of any uninterruptible power supply (UPS) system. The battery is ultimately at the heart of the UPS in terms of reliability. Understanding causes of UPS battery degradation, ways to prevent degradation and maintenance options are fundamental to ensure maximum system availability.
There are a few ways to extend the life of your data center UPS batteries and I'll talk about them here:
- Purchase the correct battery for your data center – There are many things that factor into the purchase of UPS batteries such as life expectancy, front or top access, voltage, capacity as well as other considerations. Be aware of these, for instance a VLRA (Valve Regulated Lead Acid) battery, while relatively low in cost, has low reliability. While a Flooded or Wet cell battery is much more expensive, it will give you the best reliability and a long service life.
- Is the battery manufacturer qualified by the major UPS manufacturers - Companies like Eaton/Powerware and Liebert/Emerson have qualified battery manufacturers they use for their UPS systems. These battery manufacturers include: C&D, Power Battery, GNB and Enersys.
- Is the battery fresh or surplus stock that has been sitting on a shelf - Batteries should not stay longer than six (6) months without recharging. Many times when you see a cheaper price it’s due to the need to move these off the shelf. It is difficult and labor intensive to charge pallets of them and if the batteries stay on the shelf too long they will need to be scrapped. Bottom line; make sure the da
te codes on the batteries are within six months from the date you received them. Otherwise you may be getting a five year product that is a year or more old.
- Install, manage and maintain batteries properly – Proper installation of the batteries is important and if not done correctly can lead to shortened battery life. These batteries are sensitive to room temperature, so a good maintenance practice is to keep them as close as possible to 77 degrees Fahrenheit. You need to make sure they are all kept as close to this temperature as possible so that some batteries don’t age faster than others, this can make the batteries die quicker.
- Make sure you have regular inspections of your batteries - the visual portion can unveil problems such as damaged or leaking post seals that can lead to corrosion or fires. Taking readings will tell you if one or more is not operating correctly due to a variety of reasons. A short within a cell will cause a lower float voltage on a battery. A battery with an open cell or low capacity may have a higher float voltage on it. The main problem is if one battery in a string that is wired in series (typical UPS application) starts to fail or fails it will affect the remaining batteries.
- Keep an eye on the battery's discharge status - if a battery isn’t recharged within 48 hours it can lead to damage.
Submitted by Paul Iodice, Power Electronics Supervisor - Electronic Environments Corporation
piodice@eecnet.com
Posted by Kim Otte on Wed, Jun 02, 2010 @ 09:33 AM
Building a new data center or server room is a very complex project, whether you're building a small 1,000-square-foot server room or a 100,000-square-foot data center. Here are some important facts your IT department needs to think about prior to beginning:
- Determine design requirements early in the process:
- IT hardware load projections and cabinet load densities (blade servers?).
- Electrical and HVAC redundancy options.
- Fire detection and suppression systems options.
- Site selection:
- Square foot requirement.
- Structural requirements (ground floor or upper floor loading issues).
- Proximity to utilities (power, cooling, communication).
- Comprehensive, coordinated design documents ensure a successful construction project:
- Partner with a data center consultant with specific experience in this industry.
- A design consultant with data center construction and service experience will incorporate maintainability into the design.
- Engineering site reviews during construction ensure that the design criteria are incorporated into the build-out.
- Properly document warranty and service agreements:
- Part of the construction project close-out should include warranty documentation.
- Warranties typically carry some level of parts replacement, but do not include preventive maintenance agreements.
- Service and maintenance agreements during and after the warranty period should be planned and can be budgeted at the close of the construction project.
These are just a few critical things that need to be looked at during the construction of your data center - if you don't have staff with the skill set to manage this process, it is extremely important that you hire a construction company who focuses on building data centers or to hire a data center consultant to represent them during the construction.
Submitted by Jim Stark, Design/Build Manager - Electronic Environments Corporation
Posted by Kim Otte on Thu, May 27, 2010 @ 10:43 AM
Thermal imaging or thermography is very important for different components in your data center. It is a must for all electrical and mechanical systems and should be part of the regular maintenance schedule.
Thermography allows you to detect irregularities in systems before they can become a problem. The following electrical systems should have this performed:
- Electrical switchgear
- Uninterruptible power supplies (UPS)
- Automatic transfer switches
- Server systems
- Cooling systems
There can be problems in these systems that cannot be detected with the naked eye. Using infrared thermal imaging can prevent equipment failure and unplanned outages and can help detect overloaded circuits or loose connections which can cause degradation.
Thermography also is important role in detecting hot spots in your server racks and network equipment. If these are not detected, these high temperatures can compromise your data center network reliability.
So you can see that using thermal imaging as part of your regular maintenance is really important to keep your data center or server room up and running continuously.
Have you used thermography as part of your data center maintenance? If so, tell us what problems were detected that otherwise might have ended up being a huge problem for you.
Posted by Kim Otte on Tue, Mar 23, 2010 @ 11:53 AM
If you read my blog post from November about Utility Rebates for your Data Center, then you will want to check out this blog post from AJ Howard from Energy Market Innovations (EMI) titled "Utility Sponsored Incentives for Data Center Efficiency" http://wp.me/pJn6M-Y.
It gives a little more insight into why utilities offer these incentives and why data center owners should take advantage of them.
Electronic Environments has solutions that qualify for incentives like ultrasonic humidification and room scale intelligent cooling and can help you work with the utilities in your area.
Are you tasked with saving energy in your data center or server room? This is a good way to both save energy and save money on the energy savings solutions. Let us know your thoughts on this subject.
Posted by Kim Otte on Mon, Mar 01, 2010 @ 01:50 PM
One of the topics we're frequently asked is "what are zinc whiskers?" This perplexing issue is extremely common and is found in data centers and computer rooms of all sizes.
The below post comes from Rich DeBlasi of SPEC-CLEAN (http://www.specclean.com/).
What are zinc whiskers?
Zinc whiskers are near-microscopic shavings that affect galvanized metal surfaces, usually the pedestals, stringers and th
e plenum side of floor panels. Though they can be caused by many factors, it's usually daily wear-and-tear (walking, moving equipment, removing tiles) that generates zinc whiskers. They are also quite small; all but the finest of HEPA filters aren't 100% successful in catching them.
What are the problems that zinc whiskers pose in my data center?
Although benign when undisturbed, zinc whiskers become a problem once they become airborne. They pose a threat to computers and other electrical systems in data centers. Zinc whiskers can be responsible for electrical shorts, impacting circuit impedance and can even cause disk failures, along with other catastrophes
What should I do if I suspect I have a zinc whisker problem?
Unlike some other situations, the longer the problem goes unchecked, the more extensive (and expensive) the solution is. Getting 
a lab analysis is a good start; however, hiring a data center cleaning specialist to assess the situation and make a recommendation that outlines the best method for remediating the problem is a mandatory step for any data center that is dealing with possible zinc whisker contamination.
Data center cleaning is something all facility managers need to think about. This type of service should only be performed by companies that are knowledgeable about data center environments. You don't want a traditional cleaning crew to bring in mops and soap that may cause damage to some of the infrastructure. It is recommended that a complete cleaning should happen at least once a year as part of your data center's preventive maintenance best practices.
Posted by Kim Otte on Mon, Feb 01, 2010 @ 02:50 PM
Submitted by Wally Phelps - Engineering Manager with AdaptivCool, an EEC partner
Blade servers and other high density IT equipment is becoming more mainstream by the day. Data Centers and server rooms that were designed to cool 1-3KW racks are now being asked to house and cool racks of 10KW or more. Computer Room Air Conditioning (CRAC) capacity is often more than adequate, however distributing the additional cooling air required pushes sited and the new IT gear to the breaking point.
The root cause is twofold. On the supply side there is often not enough static pressure under the floor to allow even open grate style perforated tiles to flow enough air. Typical data center design is target
ed at 0.1"wc of static pressure. Actual measurements on hundreds of existing data centers shows typical pressures of 0.02 to 0.05" wc; less than half of the design point. The causes of this low pressure are myriad, from unintended breaches in walls and support columns that go un-noticed to excess cable cutouts and simply too many perforated tiles. Underfloor obstructions also prevent static pressure from being evenly distributed as does CRAC placement that causes underfloor vortexes and low pressure from colliding air streams under floor.
On the return side, the hot air exhaust from these new high density servers must have an unimpeded low resistance path back to the CRACs; to be cooled and reprocessed. High density equipment ejects a much higher volume of warmer air than many data centers can cope with; causing other servers in the room to literally choke on the hot exhaust of its neighbors.
The solutions can often be relatively simple, such as blocking cable cutouts that waste precious cold air, adding blanking panels or moving racks to more ideal locations (when possible). Even with these measures however static pressure tends to still be much lower than design. A typical improvement in static pressure from closing cable cutouts is 20% which equates to static pressures of 0.03" to 0.06", still far below design point and not able to cope with high density racks.
A more effective solution that immediately solves poor cooling distribution is to use active airflow devices. These can be thermostaticall
y controlled underfloor air movers that install quickly and unobtrusively under existing perforated tiles. These devices deliver up to 1200 CFM to supply racks of 10-12KW with more than adequate cooling air. Even at very low static pressures! Another solution is overhead return air movers that pull hot exhaust air up and away from neighboring racks and deliver it efficiently to CRAC intakes.
The solution for cooling high density IT equipment is simple. Provide enough cool air to the intake side and ensure the hot exhaust does not affect other racks.
Posted by Kim Otte on Tue, Dec 08, 2009 @ 12:06 PM
The size and shape of a data center can impact the ability of Computer Room Air Conditioning (CRAC) systems to adequately cool the heat loads. Larger rooms will typically require a greater number of CRAC units due to limitations on air delivery of the units. So, although a single unit may have the BTU (British thermal unit) capacity to cool a given load, it may be unable to effectively deliver the cool air if the load is spread out in a large room.
Odd shapes and obstructions also present air delivery problems and can be especially problematic when trying to achieve redundancy of CRAC units. These limitations apply mostly to traditional CRAC units which deliver air through a raised-floor plenum or overhead ductwork. In-row cooling solutions can overcome some of these limitations since the units are placed close to the heat load, and redundant units can be placed in the rows, as well. Raised-floor applications with traditional CRAC units, fan-tray supply diffusers and return-air fan products can improve air delivery, redundancy, and efficiency.
Data centers or server rooms with alcoves, L shapes, or two large rooms connected in a disjointed fashion also create problems. Consistently, we find these are really two separate environments with little in common when it comes to airflow and cooling. In cases where moderate-density equipment has been placed in an alcove or other similarly obscured part of the room, some type of assisted or active airflow is usually necessary to move cooling in, move heat out, or both.
Ceiling Height 
Whether ceiling height affects cooling largely depends on how well the room is laid out. If there is a proper hot aisle/cold aisle layout with CRACs placed at the ends of the rows with at least 36-inch wide hot aisles and 48-inch wide cold aisles, a ceiling as short as 8 feet can work. It's difficult at 8 feet to use overhead cable trays, however, so a 9-feet minimum is recommended to keep data cables out of the raised floor.
If the room does not have a proper hot aisle/cold aisle layout, then more ceiling height will help promote a lower resistance path to the CRAC returns. In this case, higher ceiling heights can also be used to promote a stratification layer where hot air collects at the top. The CRAC returns can be raised up to at least 2 feet below the ceiling to collect the hottest possible air. If there is a suboptimal layout and a low ceiling, then active airflow devices can come into play to force the air away from its "lazy path" and direct it to racks or CRACs as needed.
Gauging a Room's Coolability
There are many factors to gauging a server room's coolability, but they all come down to how efficiently the right amount of cooling can be delivered from the cooling units to the rack intakes, and then how efficiently the warm air can make its way back to the cooling unit returns. The enemy at all times is the mixing of hot and cold air.
The following factors should be considered when determining a data center's coolability:
- CRAC orientation. The CRACs should be at the ends of the rows to allow cooling and heat to flow easily in and out of the rows with the minimum amount of mixing.
- Raised floor height. As a general rule, a 10,000-square-foot data center should have a minimum of 2 feet. Smaller sites can get by with less; larger sites need more.
- Underfloor obstructions. Obstructions close to CRAC units cause the most problems. As you get farther away from the CRACs, they become less intrusive. Cable trays and other large obstructions should be limited to the hot aisles if possible.
- Shape. Square or rectangular rooms will usually be easier to cool.
Floor cutouts. These should be small enough so that excess air is not leaking out to spaces that don't need cooling, if you do have them you should be using brush sealed raised floor grommets to control the leaking.
- Ceiling height. It is often difficult to get good return air paths below 9 feet.
- Ducted ceiling returns or containment. These always improve the coolability of a data center when properly engineered. Special attention must be paid to local fire code enforcement, which varies widely.
- Blanking panels. These prevent recirculation of hot air from the back to front of a rack. If there is not enough CFM (cubic feet per minute) being delivered to the front of the racks, blanking panels won't help.
What shape is your data center or server room? Have you already put some of these best practices to work and if so, have you seen better cooling in your data center?
Posted by Kim Otte on Thu, Nov 12, 2009 @ 10:20 AM
Data centers and server rooms require large amounts of energy to operate and the push now is to reduce the use of energy in data centers - green initiatives. Additional encouragement is coming from utilities throughout the United States in the form of energy efficiency rebates and incentives.
Typically, energy efficiency programs focus on new data center construction, retrofitting the current environment, or optimization of the equipment in the data center. They encourage everything from server consolidation projects to energy efficiency "refreshes," such as installing more efficient servers and cooling technology.
|
Recently Electronic Environments installed 13 Ultrasonic Humidifiers at a customer site - the customer worked with their local utility and is receiving $47,000 from the utility in the form of a rebate based on the energy savings they will receive from the ultrasonic humidifiers. |
There are many different types of incentives and rebates available from local utilities and state energy efficiency programs, so make sure you do your homework to see if your energy savings initiatives are eligible for the rebates/incentives.
According to Forrester Research - Government bodies and an increasing number of utilities have mechanisms in place that encourage business engagement in more environmentally sustainable behaviors. These mechanisms typically take on one of three forms: arbitrage, incentives, or penalties. Forrester recommends IT leaders get up to speed on the mechanisms available in their region, since the financial implications - both cost avoidance and savings - can be compelling and further improve the return on investment from green initiatives. Because incentive programs vary widely depending on your location of operation (e.g., state), a great starting point is to call your local utility provider and government representative.
The onus is on data center managers to start implementing energy efficiency measures as soon as possible. Data center power demand has risen steadily over the last 4 to 5 years - if that doesn't change, the data center industry may be headed into regulation. Different states have already enacted legislation in curtailing data center power usage, the Federal Government could be following suit. When this happens, energy rebates will disappear as well.
These resources, coupled with reduced energy bills, provide a compelling financial return on the initial efficiency investment as well as a positive message of corporate responsibility.
Check this out: Installation of Room Scale Intelligent Cooling Technology in Manhatten Bank Data Center Results in Rebate from NYSERDA
Has your company participated in any energy rebates? If so, can you share your experience?
Posted by Paul Golden on Mon, Nov 02, 2009 @ 10:58 AM
Data Centers rely heavily on power to give it the ability to perform. Testing the power in your data center is something that should be done because you need to know your backup power in your data center is strong enough to handle even the heaviest of loads.
Operating an emergency backup generator in a data center should come with the responsibility of doing preventative maintenance (PM) to ensure it is in peak condition to be able to respond to your 
critical facility's need for on-demand emergency backup power. Besides the standard PMs done on the standby generator, which includes fluid and filter changes, load bank testing is an invaluable tool to determine that a generator set is able to fully handle the load demands placed on it when you most need it.
Load bank testing allows your data center's emergency backup generator to be tested to verify its ability to run at its full rated KW output. Typically, a standby generator is running at a level far less than the units rated output capability.
Most data center customers exercise their standby generators weekly un-loaded (or no load) but very rarely test them with building loads, allowing the occasional utility loss to do the testing. For diesel engine driven standby generator sets this can cause ‘wet stacking'. Wet stacking is when unburned fuel accumulates in the engine exhaust -- this is caused by under loading an emergency generator. When exercising a generator, or running it for short duration outages, while under-loaded the engine may not reach its optimum operating temperature. When this is allowed to continue for long periods of time the unburned fuel accumulates becoming harmful to the engine's efficiency and life span.
Load bank testing of the emergency backup generator allows the engine to reach this full operating temperature and ‘burns out' the accumulation of un-burnt fuel, allowing the unit to run cleaner and more efficiently. It also offers peace of mind that your standby generator is operating as it was designed to. Any generator set, whether the prime mover is diesel or gaseous fuel driven, can benefit from undergoing a load bank test.
Load banking allows testing of the emergency backup generator with no interruption to your facility's loads. This service is in no way intrusive, and can be done at your convenience or during a scheduled PM. Most original equipment (OEM) manufacturers agree a generator set that is frequently run lightly loaded, or never loaded at all, can benefit from being load bank tested on a yearly basis.
While load banking a standby generator up to its rated output a technician is able to monitor critical engine parameters such as; oil pressure, engine temperature, fuel pressure, etc. These readings should be recorded to show that once the unit reaches full rated load it runs at normal operating temperatures and pressures.
Electronic Environments offers load bank testing and we provide a full report of the readings including generator output readings such as; voltages, amperages, KW amounts, etc.
Do you currently run a load bank test on your backup power or have ideas/questions pertaining to backup power, feel free to comment here or contact me:
Paul Golden
Standby Power Supervisor
pgolden@eecnet.com
Posted by Kim Otte on Mon, Oct 19, 2009 @ 02:44 PM
A power outage, however brief, can cause most data centers and businesses to grind to a halt. As a result, information, productivity and profit can be lost. Using an Uninterruptible Power Supply (UPS) is the easiest way to provide secure power to protect your mission critical systems. 
The UPS system protects against all types of power supply failures and disturbances. How does this work? By positioning the UPS system between incoming power supply and the load, when then the main power fails, the UPS system switches to the backup UPS batteries, ensuring an unbroken power supply is available to keep the critical load fully operational.
The need to properly size a UPS system is critical to its reliability and the systems need to be able to grow along with your data center. Data center managers need to make sure that in today's ever changing per-rack power demands - this is done correctly.
Recently Electronic Environments Corporation (EEC) Design/Build Manager, Jim Stark, offered his knowledge on the subject. I encourage you to check out the editorial at processor.com. You will find answers to questions such as:
-
What are the main factors to consider when choosing the size of a UPS for a data center?
-
What is important for readers to understand about VA ratings of UPSs?
-
Are there different considerations for blade servers vs standard servers?
-
And more...
Does anyone have additional recommendations for sizing a UPS system?