AT&T survey: Cloud computing saves cash, planet
Biz spends 69% less after move to the cloud, says cloud services provider
Putting your data in the cloud could save carbon as well as cash, says a new survey (PDF) on the advantages of the cloud from the Carbon Disclosure Project, funded by US telco AT&T. The report concluded that a large US company that made the switch now could be achieving annual savings of $12.3bn and annual carbon reductions equivalent to 200 million barrels of oil by 2020.
The findings were based on a review of 11 American companies who had already made the shift to the cloud, and compared the costs of maintaining a dedicated IT system against the costs of hosting business systems on a private cloud or on the public one.
The report estimated that businesses that transferred to the cloud spent less: only 69 per cent of their previous budgets for infrastructure, platform and software.
The savings – both carbon and financial – largely come from the reduced energy consumption of new cloud-style virtual servers as opposed to the old-school physical servers linked up to a custom IT system.
The flexibility of virtual servers means that customers can use – and pay – for what they need, meaning that there's less waste and redundant capacity. Less energy used means less CO2 is produced. Large public data centres can be more efficient than ones kept by individual firms.
Virtual servers are also much more responsive to demand – allowing new servers to be set up in a matter of minutes rather than days.
Businesses using the public cloud munch energy at a lower rate than those on a private cloud because they can be more efficient overall, the report said.
Key concerns holding back cloud adoption included reliability, security fears, and worries about vendor lock-in. Financial companies were particularly worried about data security, tech and pharmaceutical companies were particularly worried about the loss of critical IP material.
Read the full report: Cloud Computing – The IT Solution for the 21st Century here (PDF). According to its website, AT&T provides a "suite of on-demand cloud solutions, hosting services [and] cloud services: cloud computing and cloud storage". ®
What a load of tosh
Low lying cloud is called fog.
Where do cloud computing applications run?
On hardware of course, so are you saying that it is cheaper for someone to host a load of systems at a datacentre, and not at their own hardware, I just saved £3800 per year hosting on my site, using a decent internet connection.
The internet was envisaged to protect us in the case of nuclear war, nit to make us dependant upon a few international datcentres.
Keep data local where you can and "hybrid cloud" if you must use this jargon
Advantages and disadvantages
The #1 reason to use cloud computing is to make it someone else's problem when something goes wrong with the server. The #2 reason to use cloud computing is to dump your overpriced half assed IT staff that you weren't qualified to judge the qualifications of when you hired them. The #3 reason to use cloud computing is to pay less taxes because you can get government credits for having a smaller carbon footprint. It's pretty much rubbish in most cases, but if you wear a tie and have an MBA, it's a great move because getting that project going, padding your resume and leaving to move to another company before someone finds out it's hogwash can double your salary in a year.
If you're not specifically choosing a data center north of the arctic circle where the equipment can be passively cooled almost all year round, then your carbon cutting measures are nonsense, half assed solutions. Move to northern knuckland (canada) or even better use a data center in Long Year Byen, the northernmost city in the world which happens to be in Norway which would not only be a great idea, but would also protect all your international customers from privacy attacks by the U.S. government.
Almost no companies use their servers to full capacity. It's a fact. Even less companies have IT staff that understand things like queue management well enough to implement a load balancing virtual machine system to maximize server usage. Even less companies employee or can find consulting firms who can understand complex system design. The problem is, less than 1% of the so called IT experts out there actually understand how computers work. They can build them and install them, but there's barely a handful that actually understands how a microprocessor and operating system work. This is because to become an IT "expert", you pretty much only have to print a business card and talk a good game. I would be willing to gamble a months salary (as I don't believe it's a gamble) that if you took every "Qualified, Certified VMWare Engineer" and asked them to describe in as much detail how virtual machines actually work, less than one out of twenty can actually explain it beyond 'You download it and install it" without starting to just make crap up as they go along.
Using a big company like AT&T, Amazon, Google or someone else will at least provide you with a single centralized staff of people who if you're lucky will contain one or two people who actually have a clue. If you're playing numbers, you probably need to hire 100 IT workers to get two competent ones. The only "certification" I've ever seen which is worth more than toilet paper is the Cisco CCIE. The few of those I've met are generally pretty damn smart when it comes to networking. I wouldn't trust them to ever design a router and be a programmer working on one, but they do generally understand networking REALLY well when they're done with that certificate. Sadly, they're often idiot in nearly every other field of IT, but that's because they like the Ph.D.s of networking, very focused on what interests them.
If a company like AT&T were to setup a huge bank of VMWare virtual machines on a SAN somewhere, they almost definitely would know how to do things like compact them down so that the machines which are generally idling run on a small number of machines and then when someone logs in, another machine would power up if necessary and move the VM to a less crowded machine. This means that while you might have 100,000 servers in their cloud, between the hours of 6pm and 6am, nearly all the VMs can be consolidated into 1,000 machines where they can run just fine. Then 99,000 machines can be put to sleep until the following morning. If they don't have this working... DON'T USE THEM!!! This is simple planning.
Oh... one more thing about Long Year Byen.... they have some of the fastest Internet pipes (that's tubes for people who can see Russia from their porch) in the world as it's a main entry point for cross arctic fiber.
Survey commissioned by Vendor X
says Vendor X is the mostest awesome Vendor. Yay.