The fundamental problem with cloud in the current era is that hardware is actually pretty good, and has plateaued in many meaningful ways. If you're an IT manager working a 3-year depreciation/refresh cycle, you can look at your estate and think, well actually the headroom we put in at the beginning of this cycle was more than plenty, we can sweat the assets for an extra year or perhaps an entire cycle and reallocate that budget or return the money to the business. Maybe you can even avoid layoffs. Same with upgrading software, you can say version X does what we want and save the money on new licenses, training, etc etc.
Cloud robs you of the ability to manage your business in this way. You pay the vig or the lights are switched off - no ifs, no buts. I personally am a huge fan of cloud as a technology, but as a business model, it seems to heavily favour vendors over customers, in a way that makes even Oracle's licensing terms look fluffy and benign.
Yup. We've recently switched over to 4-year cycles instead of 3 for this very reason. The performance improvements are so negligible that it just isn't worth it like it used to be.
If it weren't for the bearings in cooling fans and spinning hard drives, I am certain we could push further still.
Isn't the USP of the cloud that you don't need to buy and manage the "physical" devices?
I mean yes, maybe some companies that already exist have an infrastructure management team, but if you want to start a business, the less physical stuff you have, the better.
You still need to manage them. All "cloud" saves you really is the effort of racking'n'stacking, which is 1 day out of a 1000-2000 day lifecycle.
I concede that cloud lets you get up and running very quickly - just like staying in a hotel lets you live in a foreign city very quickly too. But if you want to stay longer than a few weeks, you'd be looking at an apartment instead...
The article seems to only go into shallow depths as to the motivations for "cloud", i.e. "because microsoft want to". There are however bigger, more meaningful, and more sinister motivations for the "cloud".
Migrating products and services to be "online", which is essentially what the cloud is, is a win-win - politically and economically.
Being online means that Senior Analyst Mike Bloggs from GameCorp Co. can make monthly summaries about which ad struck, about which user did what, about how well that microtransaction performed with each user. It means that CTO Satya from Microsoft gets tax breaks et al, discussed over the meetings he often attends with government, for exchanging user data with FBI, GCHQ, etc. It means VP of Sales Matt Thebosspants from EA can control access to a game with codes, DMR-ing content.
It's a win on all sides, and there is nothing you can do about the ever-spreading cloud.
Heard of those cloud-connected toys for kids nowadays? Ain't that cute.
>"It's a win on all sides, and there is nothing you can do about the ever-spreading cloud."
Yes you can, you can run your own servers. Home servers don't have to be complicated to run, nor do they have to provide every online service ever conceived. An easy to run email server could be something that proves popular.
> An easy to run email server could be something that proves popular.
This is not able to exist anymore. While the server software itself can be automated, the never ending maintenance and defense of your IP(s) on RBLs, as well as the never ending cat & mouse game of spam fighting, is a full time job that only grows the more you use your email server.
>"the never ending maintenance and defense of your IP(s) on RBLs"
If you use IPv6 this problem goes away, right?
>"as well as the never ending cat & mouse game of spam fighting"
This can be automated in a number of different ways.
The low-tech solution is to use consensus amongst users. Allow home servers to share spam lists, and if multiple users mark an email as spam it's marked as spam for everyone. You'd still allow individual users to whitelist email domains they didn't personally want marked as spam. Similar arrangements work well for ad-block lists, I see no reason why the same approach can't be taken for spam email lists too.
The high-tech solution is to use machine learning to pick up on clues that an email should be marked as spam. This wouldn't be something that individual users would need to manage, it could come built-in with the home email server software. It could also evolve over time, both from its own research and from human guidance.
Here [1] we see that Amazon's (The Conglomerate's) profit was $513M.
In the same article, we're told:
The operating income for A.W.S. more than tripled in the quarter to $604 million.
Therefore, the conglomerate made $513M, and separately, a subsidiary company, AWS made $604M. This means that other parts of the conglomerate lost a total of $91M
Here's the 5 year old version.
If your family harvests 10 bananas, and you harvest 6 of them, and your family gives away 5 of them, you contributed more than your family kept.
I'm not an accountant, but when I want to understand corporate finance technicalities, I'll look up the definitions on investopedia.com. They do a pretty good job of putting them into "layman speak." Revenue is what you think it means[0]: 10% of all the cash Amazon took in over a given period was from AWS. Operating income is "a synonym for earnings before interest and taxes"[1].
So, in effect what they're saying is that AWS is so profitable that it has been subsidizing loss-centers in Amazon's business.
Edit: I should add that Amazon has a history of not turning a very large profit (if at all), and instead choosing to re-invest income on growth and R&D. So the behavior itself is not all that newsworthy, it more just speaks to the importance of AWS.
What it actually sounds like is that AWS is not a focus of the reinvestment in R&D Amazon is famous for. Whether it's unusual in this respect, or whether this is typical of established business units and they focus R&D on new opportunities, I don't know.
Maybe? That didn't automatically follow for me - the "more than 100%" is pretty vague. Could it not also hypothetically be that some of the AWS unit's profit gets reinvested into AWS, thus bringing down the absolute value for operating income whilst not violating the percentage relationship?
Amazon lost money for a long time and both private and public investors were fairly forgiving. More importantly, GCP is not "eating up" Amazon's revenue. Both are growing very fast because the entire market is growing very fast.
They are picking up enterprise customers left and right. The strategy there is easy.. just go up the stack and peddle solutions built on AWS. Oracle makes billions selling crap the customer never install, much less use.
Bezos speaks Wall Streets language and will get away with making little for many years.
The grocery business is a huge push that will require much broader capital investments over a huge geographic footprint. They will be back to losing money soon enough.
Whilst I don't understand how something can be 'more than 100%' of operating income, I do understand the gap between revenue and operating income being described can exist if other parts of the business are operated as loss leaders (i.e. they don't make profit on everything they sell).
Actually, no i dont think that this will happen. All it needs is one persistent, mutating virus, that can rekindle even after it has been detected once and self modify to jump habitats- sort of the malaria-melding-plague of computation, and the whole centralizing approach is prone to downfall.
Safety in numbers, safety in monoculture, always attracts this peril
My CPA wife wants a source or explanation for the sentence "A.W.S. was 10 percent of Amazon’s revenue, but more than 100 percent of the company’s operating income." Does anyone have a link?
cloud computing will be dead when organic computers come online. a single organic computer will be able to process more computations than all the computers on the planet in a thousand years.
From the article: “When has Amazon ever thought about anything other than world domination?”
From the Terminator series: Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th. In a panic, they try to pull the plug.
Cloud robs you of the ability to manage your business in this way. You pay the vig or the lights are switched off - no ifs, no buts. I personally am a huge fan of cloud as a technology, but as a business model, it seems to heavily favour vendors over customers, in a way that makes even Oracle's licensing terms look fluffy and benign.