Sounds like ultra-positive Apple style marketing.
The proof will come when the invoices appear. If they can do it, then the world will beat their way to their door.
Data backup company Acronis Storage is going to enable cloud storage service providers to undercut Amazon S3 storage pricing, according to its CEO. When the storage desk at El Reg expressed doubt, and even disbelief, based in part on Amazon's price-cutting history, Serguei Beloussov, Acronis' chief exec, said we didn't know …
- Security planning and processes to make this mildly safe
- Backup AND RECOVERY facilities
- Software engineering to create the front end (data carriers via https, sftp and webdav)
Do I bother and add legal costs to it to either write T&Cs that disclaim any risk, or some effort to keep NSA representatives out of it and protect privacy? All they need to do is push some porn peddler onto the platform and the screeching will deafen out any reason. And if they cannot win, you happen to become the target of DDoS attacks for which China will immediately get the blame. The above are actually a multiple of the hardware costs.
I'm interested in doing this right - I'm studying this right now. But the above points emerge as major upfront costs. I'm working with an architecture that is redundant in place, and replicated elsewhere - which is where you get hit with costs.
It's probably not worth the time to explain it.
It's so obvious it ASTOUNDS me how people still think public cloud is cheaper. The only time really that is the case is if your internal costs are ASTRONOMICAL. I'm thinking here massive amounts of waste etc.
I'm not going to bother doing another set a calculations here but rather just link to another data point
(from 4 years ago mind you)
http://blog.backblaze.com/2009/09/01/petabytes-on-a-budget-how-to-build-cheap-cloud-storage/
Petabyte of backblaze: $117,000
Petabyte of S3: $2,806,000
Petabyte of EMC: $2,860,000 (yes you read that right)
This is for cheap shit storage after all, it's not for transactional workloads. Though the EMC stuff could probably be used for transactional w/o an issue, there isn't mention of specifically what was priced out, I'm assuming they kept it simple and just picked a VMAX with a PB of usable storage.
Even at a small scale (for my personal server 4x2TB disks in RAID 10 in a colocation facility with 100Mbit unlimited bandwidth) it made sense. That is at a single server level with no purchasing power!
Not only does it astound me, it practically gives me migraine headaches. enough of this cloud shit please. Stop contributing to the spreading of these lies.
The first calculation is for 100TB. They assume they need to cater for 150TB of storage (RAID perhaps?) and hence to provide 100TB, they need six servers of 25TB each.
So those servers cost you a one-off $30,000 which, divided by 100,000 GB (yes I know...) and divided by 36 to get a monthly priced by GB over a three-year period (i.e. you write-off the servers after three years) gives you...
... drum roll please....
... the $0.0082 per GB per month quoted.
Really can't be bothered with the rest of the calcs... someone else can take it from here!
As pointed out, some costs are in dollars and some in euros. But that's just the start of the problems.
Some costs are monthly (rack space) and some are presumably one-off (server purchase). In that case, a pricing comparison is only valid over a period of time, which I can't see anywhere.
Without all the detail, it's just a load of nonsense.
...when a hack admits they don't have the expertise of the experts who do a particular work for a living. I love it when the online armchair experts proceed to try and tell the real experts, all about any subject. Trying to help the self-proclaimed expert is an exercise in futility because they know it ALL.
As I'm able to buy and run storage from a TLA vendor who starts with an "E" (who also isn't known for being cheap) even after factoring in power, cooling, admin costs, replicating to a different datacenter, etc for half the price of hosting the same amount of data in Amazon... I'd be willing to believe that someone else could do it as well.
In my view the article's writer has become lost and seduced by a sales person. If you need a large, permanent volume of transactional storage then the in-house route is probably going to be cheaper. You are likely to have the in-house expertise to manage that storage, access to it, backup, etc.
As I understand it, that's not the point of any AWS service. Services like (but not limited to) AWS allow an organization the ability to use large volumes when the need arises. If you need 100TB for a month or so, the arguments in this article go out the window because then you have a choice of funding a data center to cover the needs of just one month or paying the premium for being able to use existing hardware for just one month.
OK, the one month case is unlikely. But the point is, that Amazon does not seem to be in the business of replacing the data center but instead providing the ability to ramp up and down use of CPU, RAM storage, etc. on demand. If the 'on-demand' has no value and you have the skills in-house then AWS will have no value.
On the other hand, if you do not have the in-house skills or your need for a hit of CPU/RAM/Storage is short-term the on-demand route is going to look more attractive.
As ever, never listen to sales people.
Exactly - it's the difference between buying your own car because you need to drive around for years to come and hiring a taxi because you need to get somewhere today but then you are staying put - yes, the taxi is more expensive in terms of price per mile but your overall costs are much less.........
amazon has already shown that they're more than willing to price their services aggressively. so I wouldn't be surprised to see a drop on s3 prices as soon as a competitively priced alternative emerges. after all, if the numbers shown are accurate for acronis, then likely they're applicable to amazon as well (even more so given the scale of AWS).