The difficulty sometimes in comparing is, as this post points out, people don't include all the "other" costs that often dwarf the hardware costs. Power, cooling, bandwidth, etc., all cost money, and when you're not buying a lot of them, they cost a LOT of money. The cost of equipment is usually a fraction of the TCO.
To say nothing of personnel. A Netapp NAS box doesn't do you a whole lot of good unless you've got staff to manage it --- a lot of the justification for cloud offerings, particularly PaaS, is that some of those staff functions get effectively outsourced.
I found the article kind of strange. The author spends a lot of time and effort building up math that makes the case for rolling your own NAS, but then softens up when tearing that case down. Nasuni is a cloud storage company, so I'm a little confused.
It appears that the author reached the correct conclusion: the correct solution considers the requirements before arriving at a solution. IMO, the article would be much clearer if it focused on a specific use case. There are many different situations that require large amounts of storage.
If you run a video production studio, it probably makes more sense to use a local NAS like NetApp, because you're going to move a lot of data every single day. The bandwidth costs are going to eat your lunch. In that case, a local NAS makes sense.
If you run a website, cloud storage makes a lot more sense. All of your bits are going to go across the wire, so whether your NAS is local is less significant. Having the data closer to users and maintaining a high level of up-time is more important.
It's important to note that the Nasuni product (I am employee) makes cloud storage behave very much like local storage. We're not a "cloud storage" company so much as a gateway to enable companies to securely use cloud storage and have it behave like a local NAS.
Essentially the value proposition of the cloud is "we can store that for you wholesale".
If you're just a typical internet user then it's probably not worth storing data in the cloud, since hard drives/USB/Flash storage is a cheaper option, but companies can save money on the overheads of maintaining and administering very large volumes of data. For a company, storing data in the cloud is not especially safe, but then there are also risks associated with local administration too.
The main pro for storing data in the cloud. Let's say on Amazon S3 is that you also get a web server that is able to serve those same files (public or secure). Even if you need them served to millions of users. You don't have to care about scaling this thing. Good luck doing this with a NAS.
I think another key thing is in three years the price of storage will keep falling (as it has over the last X years), and with Amazon they are eating that cost (hopefully) and not sticking you with old outdated hardware.
Kind of like leasing a car, but the cars get 10x better every year
FYI -- There are cloud storage services that focus less on low-latency and more on affordable reliability/throughput for archival class data. https://spideroak.com/diy/
That benefit is also a risk though if your data in the cloud isn't encrypted. Also if most company data is stored in a highly centralised way in cloud server farms then this becomes a very conspicuous target for "cyber warfare".
> Also if most company data is stored in a highly centralised way in cloud server farms then this becomes a very conspicuous target for "cyber warfare".
Yes, AWS is a bigger target than Joe's Burgers, but the question is whether Joe will do a better job protecting his data than Amazon does.
I suspect that AWS is more secure for most values of "Joe".
