Issue 10, June/July
FOCUS CLOUD COOLING
to test for quick and wide load swings, Sorell added. A typical tester will test for a load hat is only slightly above or below the expected median.
Hypothetically, if cooling load drops from 500 tons to 100 tons within one hour, a chiller will simply shut itself off to avoid overcooling and freezing its valves. “There’s no level of automation that will fix that,” he said. “There’s no level of automation that can intervene when, all of a sudden, no chiller wants to come on.”
In terms of capacity, chillers respond to load very slowly. They can only do one thing quickly: shut down to prevent damage – something that can bring the entire data center down. Once a chiller is switched off, it may
take up to 20 minutes for it to come back to full capacity and the faster engineers try to raise capacity the bigger the chances of damage from overcooling.
A cold-aisle-containment set-up further increases the risk of a chiller suddenly switching off. In absence of the large thermal mass of an open room, cooling load fluctuations are felt even faster. “Containment makes the response to change in load immediate and so the (mechanical) equipment has to respond immediately,” Sorell said.
“The larger I can make my space, the more response is flattened out.” If power density is so high that containment is necessary, engineers would need to make the size of the contained
aisle the largest possible to prolong the amount of time a swing in load in that aisle is detected at the chiller plant. One possible way to keep a more constant thermal mass would be to use a thermal-storage system to have a large supply of cooling capacity readily available at all times, he suggested.
Answers to the new challenges cloud computing has created in cooling data centers require strategies developed and implemented by facilities and IT teams in concert. As the need to increase energy efficiency has done, so has the advent of cloud computing intensified the need to push close together the two teams that until recently have so comfortably operated independently from one another.
THE CASE FOR MONITORING CLOUD INFRASTRUCTURE
While many improvements seen in the IT industry over the past 25 years could individually be viewed as monumental, fundamental IT infrastructure has experienced only a handful of foundational “game changers.”
Today, “the cloud” – in the form of public cloud services available on the open market and private cloud infrastructure within the enterprise – will prove to be one of such improvements.
Both public and private cloud service providers focus on delivering the required amount of dynamic processing capability to meet scaling needs of their customers. But their public Cloud providers realize that their product is delivery of IT services on a per-transaction basis and their cost per transaction is a function of everything required to operate their data
centers’ hardware and applications. For them, reducing operational costs in a modern cloud requires active monitoring of all resources, including infrastructure. Such monitoring provides an accurate representation of processing, cooling and energy consumption.
A private cloud provider must calculate a safety margin above predicted and observed utilization to withstand unforeseen circumstances to avoid the traditional practice of over-provisioning. Bottom-line profitability for a private cloud provider stems from how effectively that provider is able to tune these attributes in the context of their internal guaranteed service levels and contractual obligations.
Private clouds are gaining in popularity and are becoming increasingly viewed as economically viable because they are designed to mimic
operational behavior of public clouds. Just like public clouds, private clouds are being tasked to provide low-cost IT services. For instance, demand for email services from any particular IT organization on a Monday morning will far exceed email demand on a Saturday evening, so supporting processing and cooling systems must adapt in real time.
Adjusting these systems involves adding or removing capacity.
Additionally, the IT organization must be nimble enough to adjust processing and cooling capacities to the changing demands across any number of situations with comfortable levels of headroom – all while maintaining a competitive transactional cost structure.
Mark Harris is VP of Product Management at Modius.
www.datacenterdynamics.com 33
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84