RackForce Blog
| « Calculating Your Baseline GHG Emission | Getting there is half the fun » |
November 19th, 2008 [by Doug Alder]
ICT (Information Communications Technology) is fast becoming one of the leading causes of global warming due to the enormous amount of power consumed in the production and use of ICT devices and services. Finding ways to reduce that power usage is the key to greening your IT.
These are tough economic times and for a CIO/CTO to justify changes to the CEO he or she needs to stress the economic rewards, not the technical or green aspects, for making radical changes to the company’s ICT infrastructure. As big and important an issue, not to mention being trés cool, as going green is these days, a company’s bottom line and cash flow trump it every time
Across the industry servers utilize only between 10 and 20% of their capacity. Often the same can be said for other gear, such as storage, switches and desktops/laptops (such as do your employees have both a laptop and a desktop computer – if so why?). Making better use of this infrastructure is the first step in gaining efficiency in your ICT and by doing so reduce your TCO (Total Cost of Ownership)and increase your ROI (Return on Investment).
Often in a corporate structure the answer to putting a new piece of software into use has been to dedicate a server strictly for that program. Not only is this a waste of computing resources, it is a waste of the company’s money. That server will be drawing power and require cooling 24 hours a day, seven days a week, sitting mostly unproductive and draining money from the corporate bottom line to keep it running.
Phase 1 – consolidation
The first step is to analyze your current server infrastructure and determine how much storage and computing capacity you need. Does an application really need the full resources of a dedicated server? No? Then move it onto a different server where it shares the resources with another application. Once you have trimmed this down as far as you can using traditional servers, and realized some efficiency and monetary gains from doing so, it’s time to start planning phase 2.
Phase 2 – Virtualization
New virtualization technologies, such as Microsoft’s Hyper-V, are designed to fully utilize a server’s hardware resources. Used correctly they can take hardware that is currently seeing only 10-30% utilization and ramp that up to 80% by running different services in separate virtual environments. The hypervisor (Hyper-V, VMware, Virtuozzo, Zen) shares the resources among the running services in a manner designed to equitably give each service the resources it needs when it needs it. Currently Microsoft’s Hyper-V appears to be leading the pack in its capabilities, although it’s ability to mount Linux environments is more limited than the others.
You can use this tool from Microsoft to determine what hardware you will need to move most of your data to virtual machines. Virtualization, as it currently stands and regardless of the hypervisor chosen, has two major drawbacks. First it is capable of using at most a single CPU (some hypervisors can only use a single core) and second most database programs are not optimized for virtualization (exception being MS SQL 2008 which has been optimized for Hyper-V) so in general large or busy databases should be kept on dedicated servers where you can have multiple multi-core CPUs.
By concentrating your use of server resources through virtualization you reduce the number of servers you need to maintain and that saves money through reduced power consumption, reduced cooling infrastructure needs (again saving power), less maintenance (saves on staffing costs), and fewer hardware replacements (saves capital costs.) Once you have achieved this you have saved your company money and you have, incidentally, gone a long ways towards greening your CT.
Phase 3 – Your Datacenter
Having completed the first two phases and won over the CEO and CFO it’s time to take a hard look at your data center’s physical infrastructure. Here are some important questions you need answers for.
- What is the source of power for the data center. What Is your power provider using to generate the power you buy from them: coal, diesel, nuclear, natural gas, hydroelectric, or renewable energy?
- What is the maximum heat load (watts per sq. ft.) that your HVAC/CRAC systems can accommodate without upgrades?
- If upgrades are needed, in order to increase server density, and thus the amount of heat generated, how much will that cost?
- Given the growth of your company’s IT needs
- How many servers will you need to buy in the next 3 years (assume you are using virtualization to enable maximum use of the hardware)
- What will be the TCO for that new hardware (include power, additional networking gear, upgrades, staff costs etc?)
- Will that outstrip your cooling or power capacity forcing upgrades there?
To date, datacenters have not been designed to be green. That has not been a priority and, in most locations simply impossible due to the power source. Retrofitting existing datacenters to be green, even where possible, is a very expensive undertaking as most of the IT infrastructure needs to be replaced. Even then it may not be efficient as the building itself may not have been engineered to hold the increased amount of weight required for extremely high density server clusters (where you are generating 300 to 1000W/sq. ft.)
Before you get to that point you should be seriously considering outsourcing your ICT infrastructure to a commercial datacenter that is green and which can handle those extreme loads. By doing so you will save your company a lot of money. The ability to go high density (bigger more powerful servers capable of doing more than many small servers together) on the racks means fewer racks are required, and thus less cost and less equipment for you to manage. Less equipment, both on site and remote, means fewer staff required to manage and maintain it. No datacenter upgrades means a better bottom line for your company and you as the main IT person end up looking good to your company’s financial watchdogs.
Phase 4 – Choosing the Right Datacenter Partner
Now that you have made the decision to outsource to a commercial datacenter you need to do your due diligence and choose the one that best meets your company’s needs. Here are some points to consider, can who you are considering give a positive answer to each point?
Uninterrupted Power Supply
- Fully protected by UPS systems
- UPS circuits are not oversubscribed
- UPS and all critical systems backed up by generators
- Dual utility power feeds from separate hydro electric power substations
Network Connection:
- Dedicated connection rate
- Network growth capacity
- Tier 1 upstream connections
- Multiple Tier 1 backbone providers
- Multiple upstreams BGP4 managed (multi-homed)
Quality Hardware:
It has been shown that low cost, poor quality hardware, or the use of hardware not designed for the intended application is a major cause of server failure. To avoid this, your solution should include:
- Hardware provided from a major trusted provider, such as IBM.
- Hardware designed to handle its intended application, such as a rack mount enterprise class server
This is particularly important if you are not colocating your own hardware but instead going for a sever capacity model where you rent servers from your datacenter provider.
Environment:
Datacenters have a high power to square foot consumption and as technology continues to advance this consumption rate is growing steadily. Many traditional data centers can’t keep up with the growing power demand and worse yet, many traditional data centers acquire their power from non clean energy sources including coal generators or nuclear plants. RackForce is ensuring that they are as environmentally responsible as possible. There are a lot of factors that contribute to a green datacenter, among those power source and power consumption are the two biggest.
Most traditional datacenters get their power from either coal fired generating plants or nuclear plants, some from diesel generators. None of these are green or clean. Of all the readily available current power sources, sufficient to run a large commercial datacenter (20MW+), hydro-electric power is, by far, the greenest. It produces no CO2 or any other environmental contaminant. All of RackForce’s datacenters are powered by hydro-electric energy, from the Columbia River.
Power consumption is the other area a data center can focus on to be as green and environmentally responsible. “The Green Grid Data Center Power Efficiency Metrics” created a set of standard metrics to evaluate data centers equally. One of these is PUE – Power Usage Effectiveness. Traditional data centers have a PUE ratio of 2-3 which means for every 3 watts of power brought in to the data center only 1 gets to the actual server. RackForce’s gigaCENTER has a PUE ratio of 1.38 which means only 0.38 of a watt (400% more efficient than traditional data centers) is used to power the infrastructure around the equipment. This incredible efficiency is realized through the advanced virtual technology and expertise RackForce employees across our data centers.
- Green power supply
- Actively planning and working towards carbon neutral or zero carbon status
Location:
Datacenters are subject to the hazards native to the region they are located in. Extreme weather conditions, earthquakes, and floods have all caused data center failures in the past, for datacenters located in hazardous areas. Data stored in areas where these and similarly severe hazards exist is data that is at risk. Kelowna, BC Canada, where RackForce located its datacenters, is free from those risks. The region has been rated as one of the safest and best areas in North America in which to build a datacenter. If a company’s office suffers an outage their team can relocate to RackForce’s Disaster Recovery space. Because the region is a major tourist draw it has an international airport and abundant facilities on hand to accommodate BCRS (Business Continuity and Resiliency Services or Disaster Recovery) teams. gigaCENTER offers 20,000 sq. ft. of office space to assist BCRS teams to get their operations back up and running smoothly again.
- No risk of floods
- No adverse weather related risks
- No risk of earthquakes
- Isolated Power Grid
- Temperate climate
Physical Plant:
- Stable and abundant power supply
- Extensively scalable floor space
- Raised computer room flooring capable of supporting equipment loads
- Ability to provide customers with separate secured spaces (cages or suites)
- Support for diverse high power & cooling densities (300 watts per sq/ft)
- Security and fire suppression
- Work area/office space available for BCRM (Business Continuity Resource Management [Disaster Recovery])
Phase 5 – Choosing the right services
Before you commit to colocating your existing hardware with a datacenter you need to look at that hardware’s current life cycle and determine how soon on average it will need replacing. If it is all relatively new then proceed with colocating it. However if it is getting close to replacement time then you should speak to the datacenter about the possibilities of renting new greener equipment and save on capital expenditures. For example, here at RackForce if your equipment needs do not match what we currently stock by default we can work with you and IBM to get you the equipment you need, or its equivalent. Going down this path can open up a lot of possibilities for your company to supplement existing overworked IT staff with services provided by RackForce and/or IBM that will increase your efficiency, which in turn can only help the company balance sheet.
Conclusion
Investigate all possibilities before deciding which path you’ll take. There are many factors that go into your decision and you need to consider them all if you want the end result to be greener hosting of your IT infrastructure. I hope this guide has been of some use to you in making those decisions.
Tags: CO2, datacenter, environmentally friendly, GHG emissions, GigaCenter, Global Climate Change, green grid, Hyper-V, ICT, infrastructure virtualization, RackForce, roi, tco, Virtuozzo, VMware, white paper, ZenRelated posts
Tags: CO2, datacenter, environmentally friendly, GHG emissions, GigaCenter, Global Climate Change, green grid, Hyper-V, ICT, infrastructure virtualization, RackForce, roi, tco, Virtuozzo, VMware, white paper, Zen
June 11th, 2009 at 7:36 am
[...] New post up at work Give me a little link love would ya
: [...]