We’ve slashed again the prices of Eaton Powerware Online Double Conversion Products with 40% off list price.
If you see it anywhere cheaper, let us know. We’ll match or beat any genuine offer.
We’ll be reducing other prices over the next few days so keep checking.
Just to avoid any confusion to anybody making purchases, we’re letting you know that Protx – the company we use to provide secure online credit card processing, has been swallowed up by Sage and will now be known as Sage Pay. (See Here)
We’ll be changing our text and graphics over the next few days to reflect this. Unfortunately Protx (Sage?) haven’t done a transition on this. Basically yesterday it was Protx, today it is Sage, so now we’re all left playing catch up to get our sites up to date.
Secured By Sage
Following on from my previous entry I’ve recently read that Google have lifted the veil of secrecy on their data centres and are now showing the world how they make their DC’s highly efficient. They too, have opted for the decentralised Uninterruptible Power Supply. Well, actually it’s more than that. They’ve actually incorporated the UPS into the server itself. (See this article)
What they do is basically refit the power supply with a battery, intelligent charger and some DC/DC conversion. This makes perfect sense, as you gain efficiency by removing the DC-AC inverter stage as required by all other UPS systems, raising the efficiency from low 90′s to over 99%.
The trouble with this however, is that you can’t actually post fit. You need to have your server built with this technology incorporated at the beginning, and Google custom build their own servers in any case.
There is one big drawback however, and that is they’ve completely ignored power quality. It’s all very well making systems more efficient, but to do so at the expense of power quality seems false economy to me.
An Uninterruptible Power Supply does more than provide battery backup, it should condition the utility power so that any transients, surges, harmonics and all power quality problems are eradicated before they hit your server. Google’s approach seems to ignore this and they may be leaving themselves open to power problems as a result.
I agree with the decentralised approach however, you put the UPS in, as and when needed, saving the upfront costs. Ensuring the UPS are at capacity also has the efficiency benefit. Any problems with the UPS can be easily rectified (and will only effect the server it’s attached too) and probably more importantly, the UPS makes sure that only clean power enters the server.
I read this article about an Australian University to Save £200k by a variety of energy saving schemes, one of which was to use local UPS systems instead of a central UPS system.
There are course, pro’s and con’s for each approach, but I hadn’t considered the efficiency angle before. When you look at it as a whole, then there is no way lots of individual UPS systems can be more efficient than one big one. Most centralised UPS systems will operate at no more than 50% load (due to redundancy -so if one UPS fails the other can support 100% load), and this is where a lot of efficiency is lost. Most UPS systems will be more efficient at full load than at half load (or less).
With point of use UPS systems, if you wanted to maintain redundancy, the same effect would occur. You would have two systems running at half load. Since each UPS system also requires its own onboard controller, you would think that this power loss would add up throughout the data center, in order to make the data centre less, rather than more efficient.
However, the real gain with using local systems is that you can size them exactly. With a centralised system you need to define what the maximum power consumption will be now, and at any time in the future and put in the according UPS (or opt for a modular system – but this is another blog entry). It is likely therefore, that in most early data centres, the centralised UPS are running no where near their 50% loading, whereas with local point of use systems you can just add systems as and when needed, thereby ensuring that you’re not wasting power by not having the UPS operating at its sweet spot.
We’ve actually used this approach for a customer recently. He has a small computer room, that has been built up over the years and has no overall UPS support. We’ve gone in to help and look at the options. The simplest approach seemed to be to put in a 10KVA UPS and wire this in to the existing infrastructure. This would give him the UPS support he needed. However, as his data suite was provided power by several circuits we would need to run in a new power feed. We would need to add PDU’s at the output of the UPS. We would then need to wire these into the existing circuits. All of a sudden, the actual cost of the UPS started paling into insignificance with the added installation costs.
As a result, we looked at individual UPS to fit into each rack and power the server and associated equipment individually. All of a sudden the numbers started to make sense. The KR1000J is more than enough for his servers, and occupies only 2U of rack space. So the customer has opted for individual UPS systems, saving an astounding £5,000 on a centralised system!
I’ve recently read a quite comprehensive report by the Leonardo Energy Power Quality Initiative Team, that put the costs of poor power quality in Europe at €150bn, a quite staggering figure.
Not all the costs are due to obvious power quality issues, such as blackouts, in fact, short term interruptions were the main culprit, followed by transients and surges – then blackouts. A rising phenomenon is flicker, and the costs borne here aren’t equipment damage, but rather the effects on individuals working in an environment that is prone to flicker. It’s interesting I find, as mostly power quality cost are put in the “data lost cost” and “hardware costs”, but seldom are people included in the equations.
Flicker is caused by changes in the supply waveform amplitude and is noticeable particularly with CRT’s and lighting. Workers subject to environments where flicker is a problem, complain of headaches, eyestrain and fatigue. What’s more, a lot of people are completely unaware that they have a problem.
Flicker frequencies are relatively low, but fast enough to mean that a line interactive UPS can do nothing about them. In fact, the transformer in a line interactive system may make flicker even worse. The only way to cure flicker, is either to remove equipment that is causing the flicker, remove the equipment that is showing the flicker, or fit an online double conversion UPS system. This will provide smooth power to the systems, eliminating flicker, and of course, protecting against all other power anomalies along the way.
Our product list is now so comprehensive that we feel it’s a little difficult for the novice to know exactly which UPS meets his requirements. We’ve now improved our UPS selector tool and also placed it right on the shop home page.
The biggest change we’ve introduced is that we can now identify which UPS will meet your runtime requirements for any load. Of course we’ve had to introduce a number of approximations to do this, but we feel the result is a very good guess.
We’ve also added a new feature, that will highlight the UPS which is the best price, the one that gives you the best performance, and the one that provides the best value in terms of technology and price.
Note that not all products will show up on the selector, as we need to programme in certain details of the UPS and this takes time, however we do have Powerware, Kehua and Opti-UPS all present. Although we’ve tested this, it is inevitable that we may have made a mistake or two. Please feel free to let us know if you find any.
Full version of the selector is available at UPSMart.co.uk/shop/bo-selector.php
Powered by WordPress