So you’re looking to move your servers off site. Perhaps they’re humming away in the corner of your office, gradually heating the room to a sauna like temperature. Perhaps you’ve looked at the cost of power into your server room, and are wondering just how resilient it is for how much you’re paying. Or perhaps you simply can’t get the connectivity you need, with latency affecting your performance. Whatever the reasoning, there’s 4 key deliverables to look out for when moving servers out of your building; power, cooling, network and data centre security (which has been previously covered).
Server downtime kills businesses. Technology website eweek.com posted that:
New industry research from Emerson Network Power released this week at the Uptime Institute Symposium in Santa Clara, Calif., reported that businesses lose an average of about $5,000 per minute in an outage. At that rate, $300,000 per hour is not something to dismiss lightly.
It’s commonplace too; after interviewing of 1800 businesses, techradar.com found that ‘server downtime’ was voted the primary IT issue affecting businesses – with a startling 22.5% suggesting it was their biggest concern. So with server downtime being potentially so common and disastrous, what should you look for in your potential data centre’s infrastructure?
Because server downtime is a major concern, every data centre has a resilient power infrastructure (if it doesn’t, run a mile). But the question is – how resilient? Well, unlike on-site server rooms, our UK data centre has two power feeds running onto site on a ring. These are priority A feeds which haven’t suffered an outage in over 20 years. Because they’re priority A (and also feed the local hospital), should there ever be a brown out situation, these feeds would be the first to be brought back online. These dual feeds run into two sets of parallel pairs of UPS. This data centre power infrastructure is called 2(n+n). No other data centre in the north-west can offer such resilience on the UPS infrastructure. From these UPS, two separate feeds connect to dual PDUs inside of your rack. Therefore, from the power into site, right through to your rack, there isn’t a single point of failure.
But what happens if the power into the site goes down? Good question – although we have two feeds coming into the site, we acknowledge that once the power feeds extend beyond our perimeter, they are beyond our control. Therefore we have auxiliary diesel power generators on site, with enough fuel for a minimum 50 hours continuous run-time at maximum capacity. Should there ever be a situation whereby we run on the generators for an extended period of time, we have an eight-hour SLA with a local fuel company to ensure we’re always topped up.
Power outages are a frightening possibility which can cause serious disruption. However, migrating from an on-site server room to a data centre with the correct power infrastructure can hugely reduce the risk of server downtime. With the highest levels of resilience, there’s no reason why power outages should occur. That’s why since inception in 2007, customers in TeleData with dual power feeds have never experienced one.
If your servers are in your office, you’ll know all too well how hot those things can get, which might be nice in Manchester’s climate, but isn’t doing the hardware life expectancy any good.
Differing data centres run their halls at differing temperatures. Geek.com published an article not too long ago titled; ‘Google’s most efficient data center runs at 95 degrees [Fahrenheit – equivalent to 35 degrees Celsius]’. Wired.com also has published recently how Ebay have been running their new data centre at close to 115 degrees Fahrenheit [approximately 46 degrees Celsius]. But these are outliers – multibillion dollar companies investing dramatic amounts of money into experimenting how temperature set points affect short term productivity. The findings are often mixed and contradictory, and shouldn’t be attempted without massive financial backing. Smaller data centre providers need to rely on factual, substantial reports and recommendations.
Most data centres, therefore, follow the accurate, trusted reports set out in the ASHRAE guidelines. This recommends that servers run most effectively when kept at a steady temperature of 21 degrees Celsius. Does the cooling system in your office ensure this consistent temperature without fail? If not, how much is that detracting from the efficiency of your servers? The University of Toronto conducted an interesting study in 2012 which looked into the correlation between temperature and failure rates, finding that failure rates have ‘a clear correlation with the variability in temperature’. That is to say, fluctuating temperatures are severely damaging to servers and keeping a consistent temperature is one of the most important aspects of increasing server life span.
So how do TeleData keep your servers at a consistent temperature? Our data halls use a combination of both Free Air and DX Cooling (look out for our upcoming blog explaining the difference between the two). Both halls have 7 dual compressed units working continuously to keep the temperature steady. When running at capacity, TeleData require only 5 cooling units to ensure this temperature is maintained. Therefore it is possible to lose one compressor out of each unit, and two out of two units and still maintain this consistent temperature. TeleData have n+2 redundancy on our cooling units, ensuring high levels of resilience in keeping your servers at a consistent temperature.
Of course, TeleData are continuously monitoring and regulating the data centre environment to ensure the perfect ambience is created for your servers. Our upcoming white paper, produced with Leeds University on the effectiveness of well managed cold aisle containment, documents findings into how to create the perfect ambience.
How quick is the internet access at your office? Are your download speeds capped? What happens if your one provider suffers an outage? We’re all too aware that network outages can happen. Only two weeks ago, a major Manchester based network provider offered an apology to customers who lost internet access for 8 hours. Could you lose network for 8 hours? How would this affect your business? If your servers are on, but unable to connect to the internet, they may as well be switched off or experiencing downtime. And if an hour of downtime costs $5,000 per minute as previously mentioned, that would be a huge sum of money over 8 hours.
If your servers were in TeleData, that wouldn’t need to be a concern. With 6 diverse fibre entrances into the building utilised by 16 different providers, highly resilient network is one of our fortes. All transit taken from TeleData is a multi-homed blend of at least three carriers ensuring a BGP feed which can boast an SLA of 100% on our ‘Enterprise’ model.
Alternatively, with us being a genuinely carrier-neutral data centre, you tell us your requirements and we can recommend the best 3 individual network providers to talk to. Carrier neutrality enables us to work closely with you to provide the best solution for your specific need – whether that be a high speed P2P overseas (our vast choice in networks offer quick global ping times) or a simple 10 Mbps transit.
Compare the single internet connection into your office, with the purpose built, highly redundant fibre infrastructure that TeleData can offer, and it becomes evident how much we can minimise your potential network downtime risk, whilst increasing network speed.
Alongside expanded choice and increased performance, a crucial benefit of colocating your servers within a carrier neutral data centre is that the network ecosystem that develops within the facility increases the competition between the providers themselves, which drives down the cost of connectivity quite significantly.
Speed of delivery is also greatly reduced – lead times of 90 days or more are reduced to less than a week as you are hosting your equipment at the very backbone of the internet and all that’s required to bring your service online is usually the installation of a cat5 or fibre cross connect between your kit and the closely located network core of the service provider.
Mitigate your risks
As is often the case, it’s easy to overlook potential risks before it’s too late. Your servers are valuable and, chances are, without them your business would not function properly (if at all). Consider how much money you could stand to lose if you have a server failure (whether that be from a power outage, network outage or server breakdown due to unsuitable climates). Then consider how much a data centre can minimise those risks, and ultimately how much money a data centre could stand to save you. Finally consider how not all data centres are the same. Each will claim to have redundancy, but to what level? If you’ve recognised your servers need 100% uptime, then you will want to have as many levels of redundancy as possible. And as we’re the only North West data centre who can offer 2(n+n) redundancy, why not give us a call?
Learn in more detail what makes a data centre worth working with by downloading our guide for IT service providers, ISPs and other B2B tech companies.