| |
|
 |
|
These days it can feel like data centres have a bottomless appetite for electricity. Racks are jammed with high-density equipment and new circuits seem to appear overnight. This means new power management challenges, and this month we’ve got best practices and solutions for getting power to your racks and keeping it there. We’ve also got an interview with TAB’s Jonathan Batten on what he’s hearing from data centre managers about the challenges they face in today’s business environment.
If you have any feedback, we’d love to hear it.
Regards,
TAB Canada
|
Power Management: Current Challenges
Data centre managers have been concerned with a constant supply of clean power since the first server was turned on. After all, electricity is the life-blood of the data centre, and the availability and reliability of every piece of gear in the room depends on it. But today’s data centre managers face power challenges that are dramatically different than they were even five years ago. In this article we’ll look at the issues and risks around power, and what good data centre managers are doing about it.
CHALLENGES
More Power to You
The market’s insatiable demand for computation means that data centres are using a lot more power. Increased equipment density and consolidation has resulted in a reported 10% increase in power usage in the past year and this number is expected to climb. Simply getting enough power and receptacles to the racks can be a problem.
A Shifting Load
For a long time data centre equipment drew power at a pretty consistent rate, but the power management technologies now employed in servers and networking gear have changed that. Today, managers have to deal with the fact that the power requirement at any given moment varies dramatically as computational loads and operations change, especially when blade or multi-processor servers are deployed. This can pose a significant problem when it comes to the load balancing necessary to ensure uptime.
Change is Constant
Adding to the problems of varied and increased power requirements is the constantly shifting and virtual nature of today’s data center. Moves, additions and changes can happen daily, and as they do the power, voltage requirement, redundancy requirement, and connector requirements change too.
THE RISKS
This increased power demand and varied consumption can cause tripped circuit breakers and the loss of redundancy in redundant power systems, creating overcurrent conditions which lead to:
- Equipment failure
- Corrupt data
- Downtime
- Thermal problems
- Increased service-level issues
WHAT YOU CAN DO
At TAB, we know from working closely with data centre managers that today’s power challenges require the correct mix of best practices and the right choice of equipment. We’ve compiled a list of both to help you manage your power issues.
Set the Standard
To manage power variation, establish safety margins for all branch circuits and continuously monitor them. So set your monitoring system to send alerts when the loading passes your branch-loading standard. Set the margins so that personnel have time to respond to the problem before fail. This also helps warn of impending loss of redundancy.
Keep an Eye on It
Power monitoring and management are essential. To stay ahead of the game in the 24 x 7 uptime environment, you need real time information on the power load on your infrastructure. Data centre managers need graphical user interfaces and automatic notification which report, manage, and notify based on parameters at the rack level. This remote monitoring (preferably over IP) allows you to address power issues before they become critical.
A good monitoring system combined with the right PDUs will allow you to:
- Receive alerts remotely when power or environmental conditions exceed thresholds
- Control power at the outlet level, along with pre-engineered and easy to use tools to monitor and control large numbers of outlets.
- Monitor input current and determine if there is available power to plug in other devices
- Accommodate diverse power options
- Remotely reboot servers and other network devices
- Sequence power during start up to avoid power in-rush
- Monitor temperature and relative humidity in data center cabinets
- Power off at the receptacle level so that new equipment can be properly installed
- Appropriately size power infrastructure for future projects
- Allow for predictive failure analysis
Manage the Proliferation of Branch Circuits
Increasing the power distribution to the rack will help eliminate the need for re-wiring as well as provide local and remote monitoring.
Stay Flexible and Avoid the Hot Work
The future is adaptability. Your rack power infrastructure should allow quick and tool-less changeover for different voltages, power capacities, outlets, and DC.
It should provide enough power to supply the maximum anticipated load to any enclosure at any time without re-engineering the power system, and have the ability to add additional circuits as needed. And a good rack enclosure power system should be adaptable to as many combinations of equipment as possible without having to perform “hot work”, which requires the specialized knowledge of an electrician, is costly, and if it isn’t done properly poses a major risk to availability.
Redundancy: Worth Repeating
Power redundancy (or fault tolerance) at the enclosure level is essential. A good way to increase and ensure availability is to provide two independent power supplies through power strips to the equipment in your rack that can support dual power feeds. Remember, for this to work effectively the loading of breakers within each power path must always be less than 50% of trip rating during normal conditions. If the equipment is not designed for dual power, switches should be employed.
Simplify
The average data centre is rammed full of power circuits. Minimize the number of breakers and amount of wiring between the UPS and the loads, cutting down on the number of different types of power distribution provided, and/or standardizing on a single power feed to your racks increases your flexibility and minimizes human error, a big contributor to power related downtime.
Regular Power Audit and Maintenance
Anything you do in the room, from installing new systems, performing upgrades, and physical layout changes can have unintended consequences. Conducting regular power audits and scheduled maintenance will help you avoid surprises.
Cord Location, Location, Location
Addressing power issues means deploying power cords and strips. Power strips are best located in the rear of the rack for easy accessibility and identification. They should be mounted vertically so they don’t take up valuable U space, block the air, or force all the power cords to route to one location. And this way the receptacles are in line with the power cords of the equipment in the rack. Remember, use shorter power cables for better cable management.
Spread the Load
If you find you’ve got enclosures where the load exceeds the design average value, splitting the equipment among multiple rack enclosures will distribute the load more evenly. This is an especially good strategy for those not constrained by space.
Talk To Us
All indications are that the two biggest drivers behind power challenges in the data centre, both varied usage and the demand for more and more power, are only going to have a greater effect in the next few years. Staying on top of this is critical to maximizing uptime, and TAB is here to help.
Whether your issue is space, power, cooling, or lack of infrastructure, we specialize in pairing expert advice with the right product portfolio. For more information or a free space analysis, contact us at 800.667.4020 or visit us on the web at www.tab.ca and click the "Technical Environment Solutions" tab.
Featured Product: Paramount Powered By PDI
Increase the reliability of your IT systems with Wright Line’s 80kW in-row power enclosure, Paramount… powered by PDI, because today’s expanding data centres demand fault tolerant, redundant and scalable branch power distribution solutions without the cost and complexity.
To learn more about the Paramount enclosure or the available accessories,
contact TAB today.

Key Elements:
Dual, primary and secondary power feeds satisfy system redundancy and maintenance requirements. Simply connect to any standard subfeed circuit breaker of a PDU or 208V distribution board.
Freestanding and self-contained, the in-row power enclosure allows quick deployment and provides increased branch circuits on the data center floor.
Pre-engineered and factory tested, it eliminates costly electrical work while reducing field failures.
Cut-to-length, fully terminated whips offer high flexibility while increasing the speed of deployment for the multiplicity of IT equipment.
The in-row power enclosure and Overhead Cable Management System reduces the length and amount of cables between the PDU and IT equipment, ensures code compliant distribution and prevents underfloor obstructions.
Expand branch circuits incrementally and on-the-fly throughout the data center without needing to power down any of your critical equipment loads.
Remote and local power measurement of each individual branch. Communication with building control via Modbus and SNMP allows trending and alarm thresholds providing you the central vision and control you need.
To learn more about the Paramount enclosure powered by PDI click here to download a brochure in PDF format.
TAB Interview: Jonathan Batten, National Sales Manager, TAB Canada
Recently the Quality Data Centre caught up with Jonathan Batten, National Sales Manager for TAB, and we asked him about the challenges facing data centre managers today.
QDC: Your work takes you into a lot of data centres and you spend a lot of time with the people that run them. What are you hearing from data centre managers?
JB: Well, I can tell you that if you thrive on challenge in a high pressure environment, now’s the time to be running a data centre. There is a tremendous amount of data processing going on out there, and every manager I talk to – whether they run a massive server farm or just have a couple enclosures – is telling me that they’ve never been busier.
QDC: Why is that? What kind of challenges are you hearing about?
JB: Well I think the biggest issue comes from the fact that technology is changing at such a rapid rate and this has a big impact on IT infrastructures. I mean, we’ve gone from big iron to blade servers in a relatively short time. Trying to get these new technologies, the blade servers and storage devices, into the existing data centre structures is a significant challenge.
QDC: So staying ahead of the technology curve is the biggest factor?
JB: It is more than that. These days the pressure from the executive level to respond to business changes in increasingly shorter timeframes is also a big one. Businesses have to be leaner and more responsive than ever before, and this gets pushed down to the IT level. Projects that used to take a year are now having to be completed in half or even a third of that time.
QDC: What does this mean for the day-to-day operation of the data centre?
JB: Well, the data centre has to be above all a stable environment and here you have these factors driving rapid change and instability: new technology and business pressures. The managers I talk to are busy with the balancing act of staying adaptable and flexible and still making 24 x 7 uptime a reality. All of this while having to keep costs down, I might add.
QDC: What about the traditional pain points in the data centre, your cable management, thermal issues, and power management? How do today’s pressures affect them?
JB: Still very much there. As you might expect, the constant change and growth has only exacerbated the problems we usually see. Remember, the equipment that is being deployed is becoming more dense and powerful all the time. And there is simply more gear being deployed to meet the demand for data processing. These server rooms are just eating electricity. Most of the managers I talk to describe these new servers as “insatiable”. So of course they way they are managing power has had to change. You’ve got to get more of it to the rack level, and you’ve got to monitor it closely. Same goes for heat. Anything using that much power is going to give off a lot of heat, and everyone has had to re-think their cooling strategy. A lot of data centres have outgrown their original cooling infrastructure and finding ways to improve cooling in the enclosures is key. Cable management is a challenge too because with all this equipment that is constantly changing you’ve got more cables and electrical outlets that need to be properly dealt with so that all these moves adds and changes can be made without dropping the system or losing availability.
QDC: What kind of solutions are typically being employed to deal with these new challenges?
JB: When it comes to data centre management there certainly isn’t a set of rules that tells you how to handle every situation. What we see in our countless visits to data centres is an ever-evolving set of best practices and we do our best to share that experience with our customers. On a general level I’d say that the most managers are focusing on a few key solution watchwords now: standardization, compatibility, scalability and flexibility. Most of what is happening is happening at the rack level, and we encourage people to standardize on one enclosure that can handle all your equipment and can grow with you. The products that are out there now have come a long way. The new in-row power enclosures – racks where the power source is actually located at the row level – are a major advancement on the power management front, for example. This technology is going to save people time and money because they are designed to be able to quickly add and swap out equipment without having to call an electrician or losing any availability. It’s a cliché but really, standardizing on the right enclosure for your needs goes a long, long way to maximizing your efficiency and uptime.
QDC: Any emerging trends you see coming on line in 2006 for the data centre?
JB: One thing we are seeing more of is the “building on demand” strategy. A lot of people got burned by building big before, and you end up with a lot of expensive space and equipment that can’t change as you need it. So what is happening is that people are looking for vendors that can provide them what they need, whether it is enclosures or PDUs or whatever, as they need them. Especially co-location sites. They want to be able to call you up and say hey we need another couple of racks in here ASAP and get it done. This is one area where TAB really excels. We’ve got a great relationship with our suppliers and our response time is amazing. It’s moving from traditional client-vendor relationship to more of a partnership. Companies want to work with someone that can look down the road and anticipate what their needs and configure solutions to accommodate that. I think our ability to do that is a major contributor to our success.
QDC: What would you to say to managers who are out there looking for solutions to any of these problems right now?
JB: The best advice I can give is go with someone who can provide you an end-to-end solution. They’ve got enough to do without having to think of everything that is happening at an infrastructure level. A company like TAB has years of experience and knows how to fix the problem. There are a lot of people out there slapping on band-aid solutions that come back to haunt them down the road, both from a cost and an uptime perspective. Like everything else, cost and efficiency in the data centre are paramount, and why not go with a vendor that can combine the right equipment and knowledge into a long-term plan to address both of those?
About Jonathan Batten:
Jonathan Batten is the National Sales Manager for TAB and has been with the company for almost a decade. Jonathan’s involvement in countless projects in every major city from coast to coast has given him a deep knowledge of the data centre and the solutions they require.
Send this interview to a Friend
|
 |
|
According to IDC, the worldwide server market grew at an 8.1 percent year-over-year rate in the third quarter of 2005.
|
|
|