ns
practice that IT staff make a well informed technology decision, only for their request to be subverted by the purchasing department, with the rationale that technology today is built to support specified standards and therefore all solutions are equal. If only life was this easy. Take, as an example, a company that
deployed a large core networking switch with a fully loaded weight of 200kg. The purchasing department believed that this weight was the key factor and looked for the cheapest cabinet which could be deployed in a hot aisle / cold aisle configuration. In this case, however, when the cabinet arrived and the switch deployed, the patch cords that were needed with cross connect panels, more than doubled the weight of the switch. Apart from the struggle to get 1000 patch leads into the space, there was no apparent problem with the cabinet. However, overloading racks or cabinets has a dramatic effect over time. We know from our own experience that the cabinets can start to twist and this breaks the seals for the air containment. Staff will suffer the inconvenience of doors that don’t fit properly, but there is a bigger price to pay for the company because the gaps allow the recirculation of the air, with hot air and cold air mixing which reduces the computer room Air Conditioning unit’s efficiency. Whilst the purchasing process saved a few pounds, the Operational Expenses took a long-term hit, incurring higher cooling expenses.
Cost of Ownership If Data Centres are to survive as affordable resources, they will have to accommodate the effects of Moore’s Law. IT staff need to have the planned time to keep up to date with technology progress and even more time to plan the adoption of new technology and to evaluate its relevance to the organisation. This means developing a cost justification based on power savings, cooling savings and maintenance savings against deployment costs, that is, the Total Cost of Ownership. The disruption comes from the ever increasing availability of Infrastructure
as a Service, such as those introduced by IBM. The pricing model offered would make any CFO question why his internal resources cost more to run, than renting infrastructure with a stricter set of SLA’s. No IT department is going to survive against these financials, supporting a traditional Data Centre construction. To survive this change, we need to manage the ever-escalating Operational Expenses; that is, move away from the product, price and discount to understand the ‘real’ costs. The purchasing decision should rest
not only on what a piece of equipment costs to deploy, but also what it costs to maintain and how it helps organisations to handle the change both now and in the future. For example, a simple solution to cable pathways is a wire basket; a piece of equipment that most purchasing departments believe is only differentiated by price. However, Forrester Research has examined this very problem and found that the time taken to deploy a wire basket differed from one manufacturer to another by as much as 75 percent. In real terms; with one solution it could take four days, with another it would take one day, equating to a massive difference in deployment costs.
What about handling the future?
Cable deployments are moving away from onsite deployments to factory pre-terminated solutions, and moving to SFP+ cabling schemes , or parallel optics for 40GB or 100GB which are not available for on-site assembly and need specialist equipment. Only one of the cable tray manufacturers at the time of the testing by Forrester Research could accommodate the pre-terminated cassette construction. It’s the short term purchasing
approach that has resulted in IT budget being soaked up into Operational Expenses so we need to break the purchasing grip of buying the cheapest product and go for the most cost- effective long term costs. An example of this can be seen in the advancement of switches since IEEE802.3az - the ‘Power Down while Idle’ specification, or PDI - which means that switches auto sense when ports are idle and closes them down. For a data centre with 1000 servers, today’s 10GB Ethernet
ports consume 5.5 watts on copper. For resilience, there will be an ‘A’ network and a ‘B’ network with 2000 switch ports and 2000 sever ports. That equates to 22 kilowatts of power, just to maintain the ports. It is estimated that the PDI will save up to 75 percent of the Switch power, or 16.5 kilo watts of switch power plus another 16.5 kilo watts cooling power. What is the long term costs savings of 33 kilo watts? IT should have the time to answer these questions.
Conclusion This year holds significant challenges for anyone involved in the Data Centre business from design to operations. The world is moving to cloud computing, and virtualized servers with spare capacity to meet the on-demand requirements for business applications. As organisations work through these changes of Consolidation, Virtualization and Automation, the business is expecting a new services paradigm. They want instant IT gratification, always available at the lower cost. This is deliverable if the IT staff are allowed to do the task they have trained for and focus on the deployment of solutions that have real value to contribute to efficiency and long terms savings, rather than those which offer the biggest discount. We sent men to the moon on the equipment with the lowest bid we do not want to send men to Mars with equipment that uses too many resources, cannot be maintained effectively and cannot be adapted to the changes they will find. With this long term approach in mind, CTO’s should take this opportunity to redress the balance of power so that the correct decisions are made by IT Professionals.
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