Bank of America Says It Saves $2 Billion Per Year By Ignoring Amazon and Microsoft and Building Its Own Cloud Instead (businessinsider.com) 128
Bank of America has bucked the Wall Street trend by building its own private cloud software rather than outsourcing to companies like Amazon, Microsoft, and Google. From a report: The investment, including a $350 million charge in 2017, hasn't been cheap, but it has had a striking payoff, CEO Brian Moynihan said during the company's third-quarter earnings call. He said the decision helped reduce the firm's servers to 70,000 from 200,000 and its data centers to 23 from 60, and it has resulted in $2 billion in annual infrastructure savings.
Paywalled Article... (Score:3)
SO I can't RTFA BUT the title and summary don't make sense..
If you're in the cloud you have VMs not servers and you don't have any data centers (well at least the stuff that we're talking about isn't in your own data centers) SO how can they possibly say they save $2B compared to Amazon/Google/Oracle/etc when all of the tech comparisons are only vs their own internal resources.
As far as I can tell (unless the article has a BUNCH better detail) all they did is reduce their internal footprint by virtualizing (aka private cloud) so as usual bad title/summary?
Re: (Score:2)
It's simple.
Instead of paying for time on other people's computers, they pay for time on their own computers. Which used to be called "owning a computer," but now we call it "the cloud." So instead of paying cloud providers for cloud computing, they are cloud computing on a cloud provider that is themselves.
HTH.
Re: (Score:1)
A "private cloud" is not just owning a computer obviously. I see you're joking, but frankly, the cloud is a distinction b/c CPU cycles are commodotized, so instead of allocating them to sets of services on a whole machine basis, all the services can basically share the reserves in case of a spike in usage.
Re: (Score:2)
I bet your great at parties.
Anyway, that's not new and it used to be called owning a computer. Usually a mainframe system.
Another 20 years and the PC and Could might have all the feature of a mainframe from 1980
Re: (Score:2)
"Bs. You could self-provision multiple vmâ(TM)s on your mainframe?"
Sure you could, at least in VMS. IIRC, on our smallish 4-CPU system they sometimes ran 1 VM for production while the other three were doing testing, each with a different date/time or OS version. I don't know about other mainframe systems, but VMS was quite flexible in its capabilities.
Re: (Score:2)
A "private cloud" is not just owning a computer obviously. I see you're joking, but frankly, the cloud is a distinction b/c CPU cycles are commodotized, so instead of allocating them to sets of services on a whole machine basis, all the services can basically share the reserves in case of a spike in usage.
Oh, I see. So you're saying they run software to manage their infrastructure? That is pioneering.
Re: (Score:2)
It sounds like a company who subcontracts to itself.
"Hey Johnson, can any of our employees do this?"
"No, but we can hire ourselves as a contractor, and then some of our employees will remember how to do it."
Re:Paywalled Article... (Score:5, Insightful)
Re: (Score:2)
What are you all on about? Having a private cloud is perfectly acceptable and normal. It means that instead of running servers, storage and network in department and application specific environments, they run a large hypervising stack to mesh all their compute needs together in a consolidated form. Private Clouds are not new.
Yep. The company I previously worked for was doing cloud computing before cloud was cool. Owned multiple HP Blade server clusters running VMWare with fiber attached EMC^2 SAN hardware. The only bad part at the time was the extra cooling those blade servers needed. They put out a ton of heat, especially the HP C-class blades!
Ran virtual Windows servers on top of those VMWare hosts hosting Citrix virtual servers on top of the Windows hosts! A VM within a VM. Worked very well. And this was back in the e
Re: (Score:2)
Re: (Score:2)
Basically it's a classic file and systems server network with services attached (like RDP, DFS, etc.) that create an internal cloud. This has been doable for a very long time, it's just that Google, Amazon and Microsoft have built general purpose clouds that can handle multiple companies. It certainly isn't magic, and yes, it's likely going to mean hiring contractors to build and maintain those services, but what of it? I work for a government contractor, and we have access to parts of the Government cloud.
Re: (Score:2)
The difference between a private cloud and just virtualizing or running on your own servers; Is in the consumption model and in the IT management processes. In a cloud environment you have a system which provides APIs offering services consumed by others rather than a single unit just manually doing things like clicking a "Create/Clone VM" button and deploying a server... "Building your own cloud" means scrapping the previous generation of misguided overhyped practices such as ITIL/ITSM, and using
Re:Paywalled Article... (Score:5, Informative)
Because they're intentionally misinterpreting the use of the term, "cloud" as it applies to Amazon Web Services and other providers.
What they've done is use the buzzword in their own way. They've constructed their own datacenters, employed their own IT staff, but probably used the same sorts of virtualization technologies that these cloud providers use. And on that vast a scale it probably does save them money compared to paying a third-party, especially given they'd still have to employ an army of IT staff to do proper security.
Re: (Score:2)
It's the correct use of the term "cloud". CPU cycles are commodized at a fine granularity instead of on machine increments. We used to waste a lot of CPU isolating services on diffrent physical machines. It was pain to move them to a bigger one... now you can easily do it as an operator of the "cloud". You can also put desktops in this cloud, and have fairly dumb terminals to log in. The fine granularity of resource sharing saves tons of money. Not paying someone else to host the actual machine and be
Re: (Score:3)
Indeed. It's pretty obvious that paying for IT services is going to be cheaper than paying for those same services from someone else, plus administrative overhead, plus profit.
Of course that's assuming your demand is fairly stable - if it can bounce around dramatically from week to week then it might make more sense to rent it on demand, from someone renting to enough other customers than random variation averages out.
You also need to be operating at a large enough scale that you don't have underworked high
3rd party cloud can jack up rates and hold you hos (Score:2)
3rd party cloud can jack up rates and hold you hostage don't want to pay well we will wipe all data in 60 days after you stop paying.
Own for base load, cloud for bursting... (Score:5, Insightful)
BofA isn't the only one to have figured this out. It's common to own your servers for the base load that you need (somewhere around 90 to 110% of base) and then use someone's data center for bursting capability. If you can run your datacenter at nearly full utilization it's cheaper to do it yourself (once you reach enough scale) and then if you need extra capacity buy it from a cloud provider.
The same is true with lots of production equipment -- if you mostly need 10 trucks in your fleet, you should own 10 and then rent for the times that you need more. Yes, the cost for the 11th truck will be higher, but you only pay for it when you need it.
This is sort of business 101.
There is no "cloud", just other people's computers. At the low end, it's worth paying someone to manage the facilities. At the scale of BofA you can do it yourself.
Back in my day, we called it "timesharing". Now you kids get off my lawn.
Re: (Score:2)
"There is no "cloud", just other people's computers."
So you are clueless then?
IT's not even close to timesharing. JFC, gray beards like you make the rest of us gray beards look like idiots, knock it of.
Re: (Score:2)
No.
It's where the front end doesn't know where, physically, the data is on the internet.
It literally comes from the cloud drawing the represented the internet.
Post 2006.
Pre 2006 it just meant doing business on the internet,, pioneered in 1996.
In one of life's little twist and quirks, I was at both events.
Users don't know machines. Public and private (Score:4, Insightful)
On a cloud infrastructure, a user can spin up an instance of Centos 7 without knowing or caring which hardware their instance runs on. They have to worry about which drive the files are stored on - there is a SAN that stores the stuff, and it's automated.
There are public clouds and private clouds. On public clouds like AWS, a developer can spin up a machine for testing. They don't have to have somebody from another department build theitlr server with a motherboard, CPU, etc. A developer working for any company can do this (hence public).
On private clouds, a developer can spin up a machine for testing. They don't have to have somebody from another department build theitlr server with a motherboard, CPU, etc. Only a developer working for a particular company is allowed to spin up servers on this cloud (hence private).
Who "builds servers?" (Score:2)
They don't have to have somebody from another department build theitlr server with a motherboard, CPU, etc.
Huh? I get your point, but what companies have employees building servers? I've worked for huge places , including Bank of America, down to small, broke, non-profits and servers are ordered from major vendors. No one is in a back room with bins of screws and empty chassis waiting for orders. I am pretty confident most administrators have no clue what motherboard their machine is running, just the basics you'd see in a system monitor, like RAM and CPU model.
When I suggested that very idea 20 years ag
Re: (Score:2)
You make a good point, most companies only build servers on Dell's website, choosing a CPU, amount of RAM, storage type and amount, etc by clicking. Most don't screw them together initially (though they may install more RAM or storage later). I think what I actually said was a department choosing CPUs, etc. I don't think I mentioned a screwdriver. :)
Funny thing, though, the office next to mine is full of hard drves, chassis, etc spread across the desks. My "best friend" in the office is commonly seen wit
Re: (Score:2)
I disagree (Score:2)
While the term cloud has been used for all sort of things... fact is, the industry has a term for "private cloud" which is privately owned virtualization. And as I said, it's the realization of a goal of having fine grain commoditizing of CPU and other basic resources. Terms change, but as far as I'm concerned this is what the meaning of cloud became, and I remember computing presentation in the 80s charting the way it would happen. It's certainly descendant from the idea of a cluster.
but why would we wan
Re: (Score:2)
Cloud is running software(apps, data, etc.) over the internet.
Literally that is what it means since 2006. Prior it meant doing business over the internet. YOU know, the cloud drawing.
Re: (Score:2)
I would say clusters are a term that tend to suggest batch scheduled workload.
'Cloud' is providing endusers the ability to self-serve without any human intervention from a higher authority org.
So a vmware cluster generally required a department to ask the admins to create and then a human would do that.
'Cloud' refers to when admins are able to define more self-enabled mechanisms for departments to get new things carved up, often hooked up to internal accounting such that people don't leave dead instances ar
Re: (Score:2)
"especially given they'd still have to employ an army of IT staff to do proper security."
And I'll guarantee that they're **NOT** doing anything like "proper security" or having an army of staff on hand. Bankers are only people in the world cheaper than lawyers. They'll subcontract the work to the lowest bidder, as always, and then when there's a major security breach they'll pretend to be shocked, as always. Within a few weeks their stock value will have recovered, which is all the executives care about.
Re:Paywalled Article... (Score:5, Insightful)
Re: (Score:3)
This is just informing the unknowing general population about BoA's use of private cloud as opposed to public. To your point, this is neither news nor stuff that matters; this is a meaningless PR piece.
Re: (Score:2)
"the bank estimates its private cloud is 25% to 30% cheaper than public providers,"
That is a big deal to us nerds who deal with executives in decision making.
I'm sorry you work in a tiny cube and have no actual input so article like that have no value to you; but stop projecting what is news for nerd onto those of us actually helping make billion dollar decisions.
Re: (Score:2)
It provides a nice counterpoinnt to the overwhelming marketing material stating that "everything is in AWS or Azure by now, why aren't you?"
Re: (Score:2)
Re: (Score:2)
You'd think with a low id on /. that you'd remember that when Unix was all the rage, most users never even knew where the data centre that the box was located. How is that really different from "the cloud?" The only difference is Amazon and MS let you spin up virtual machines and supporting tools like firewalls, and network routing without an Amazon or MS admin.
The biggest differences are short-lived instances and PAYG billing instead of renting servers or even virtual servers and then having to keep them utilized.
Re: (Score:2)
No. This is understanding economics. In order for someone to provide a service they have the cost plus the profit markup. Nobody opens up a business and gives away services at a loss. If they do they close their doors. It's just the perception that it is cheaper. Most can't even say what their requirements are. This is the same as people comparing the rental of a 500 square foot apartment to a 2000 square foot house and coming to the conclusion that renting is cheaper.
Cloud is expensive. You pay for the har
Re: (Score:2)
Nobody is dumb enough to think this is magic but anyone who thinks they're going to actually do this cheaper "on their own" is probably mistaken.
If you are only using a fractional piece of one machine or you need geographic zone redundancy but you are a small business with no reasonable access to the requisite real estate, this may be correct (though who says you need to have super high end equipment and there is a high chance your small business doesn't need that level of redundancy).
If you are a large institution that can keep thousands of servers well utilized and you have real estate in multiple geographic regions anyway, there is a rather larg
Re:Paywalled Article... (Score:5, Insightful)
It's not a great title or summary. In essence, BoA has made a major investment in private cloud consolidation. It's reduced their data center footprint massively and is saving them a ton of money. Their CEO is next looking at third-party providers, saying they don't need to own the hardware, just find someone who can do the job and meet their requirements. So this is all part of a staged effort, rather than a master plan of avoiding public cloud offerings.
To your point, Amazon AWS and Microsoft Azure actually do cost more money past certain scales. You get some benefits there (only paying for what you use), but ultimately you can do it much cheaper if you do it right if you're past the break-even point. Want an example? Store 500TB of data for me, rolling over the course of 6 months, pulling about 50TB per month of that back. Price what that looks like in Azure vs a colo setup. Include 3-year refresh cycles on all hardware. 10% growth in data year-over-year. Look at a 5-year and 10-year average per-year cost of each option. Look on in amazement as you find that the cheapest way of doing it in Azure or AWS is 3-5x more expensive than doing it yourself.
you also pay to recvie space (Score:2)
you also pay to have reservations.
Re: (Score:2)
That is a big part of it, but there is a lot at the basic end. It has been a work in progress for over 15 years, before “cloud” was real, and well after virtualization made a huge dent.
I haven’t dealt with them in about a decade, but they were really trying to eliminate and consolidate on the application side and streamline business process. I am amazed they still have over 20 critical data centers though; they were projecting to be down closer to 10-12 by now.
Re: (Score:2)
Re: (Score:2)
Still the summary is either false, or the people that made the statement were idiots.
The cloud is still computers, you've just moved your data from your hardware to somebody elses.
Of course it says they built their own software, but unless they're using their own machines, it's still on somebody elses boxes.
Heavily optimized software can reduce hardware needs, but no software in the universe can magically reduce the required hardware by about 65
Re: (Score:2)
NO full video console access at the big ones and y (Score:2)
NO full video console access at the big ones and you need to fit into there sizes as well.
Re: (Score:2)
This is great news... (Score:4, Funny)
... for people that Bank at Bank of America b/c they are sure to pass that saving along to you, their loyal customers.
Re: (Score:3)
Re: (Score:2)
You jest, but ultimately it does. BofA gets a little better at delivering services more cheaply, so they are able to undercut others by a little, or other companies do the same and then BofA has to offer cheaper/better services to compete.
Most of it goes into profits for shareholders, of course, but it does ultimately benefit customers. Certainly more than e.g. governments being more efficient, they don't cut your taxes when that happens they just find other (mis)uses for the money. In fact, there's no real
Re: (Score:2)
... for people that Bank at Bank of America b/c they are sure to pass that saving along to you, their loyal customers.
You misspelled "shareholders" -- who will probably see some actual earnings from this.
How about BoA makes a public cloud offering? (Score:1)
Maybe they should get into the cloud business? Google, Amazon, and Microsoft are making hand over fist.
Re: (Score:2)
Correction: Making money hand over fist.
If BoA could make something that is designed from the ground up for PCI-DSS, other financial institutions would jump on it in no time.
Re: (Score:2)
eh you would think so but... nope.
Being better or specifically designed for it is not enough.
You still need connections and a presence. Why do you think sales guys are still a thing despite the fact that we have the internet and you can show your business portfolio online without a sales guy?
it is still an organic world no matter how you slice the pie.
Re: (Score:2)
Maybe they should get into the cloud business? Google, Amazon, and Microsoft are making hand over fist.
Correction: Making money hand over fist.
You're probably correct both times. :-)
Re: (Score:2)
PCI-DSS is a framework that comes from within a company and is basically a number of common sense security protocols. Nobody wants to make the claim they'll make your business PCI-DSS compliant, the standard is too broad and vague, conflicts with itself in places and requires everyone in the company to cooperate to work towards an interpretation of it.
Same with HIPAA, Amazon, Microsoft and Google don't sell you any HIPAA compliant system, they'll tell you their systems can be made compliant but put the onus
No surprise (Score:2)
Buying cloud services is pretty expensive. It basically is worthwhile only if you are not large enough to justify your own cloud operations team.
Re: (Score:2)
Depends on a lot of things. Sometimes what you need to do is bespoke enough that it warrants investment inhouse. You don't have to be big enough, maybe just unique enough.
Re: (Score:2)
That is the second valid reason to do it yourself. Quite correct.
They saved by moving off of AIX/IBM (Score:2)
They saved $2bn of their money by moving off of AIX and IBM hardware. If they had gone to public clouds they would have saved more.
Re:They saved by moving off of AIX/IBM (Score:4, Interesting)
Nobody at BoA's size and scale saves money going to public clouds. Anyone who thinks they do hasn't done it or even properly investigated it.
Re: (Score:3)
I used to know the BofA/Nations architecture and implementation processes. They used one server per app, making their infrastructure underutilized and expensive.
Given their infrastructure, they could have saved hundreds of millions just using VMWare, except they can't because you can't virtualize AIX. And you can't cloud-ie AIX either.
Like I said, they probably saved their $2bn by buying less servers and consolidating, which is what they probably mean by "building their own cloud."
one server per app works good in an VM cluster (Score:2)
one server per app works good in an VM cluster setup.
Re: (Score:2)
That was before closing all the California data centers and the big expansion in Villa Park, KC, and Richardson. Did you know Carolyn R... I remember her saying they could move all 1455 resources in less than a month.
Re: (Score:2)
You can virtualise AIX just fine. AIX can run in an ePAPR VM (a standard for virtualised POWER/PowerPC instances). IBM supports running AIX (and Linux for that matter) in QEMU ePAPR containers on their pSeries hardware or whatever they call it now.
Re: (Score:2)
If you're big enough, then roll-your-own cloud services will almost certainly be cheaper than buying those same services from someone else, plus paying for their administrative overhead, plus their profit margins.
The key is, you have to be big enough that the economies of scale of running a massive data center kick in. If you're a small business whose computing resources can be satisfied by a few hundred servers... you're probably nowhere close to that.
Re: (Score:2)
If you're a small business whose computing resources can be satisfied by a few hundred servers... you're probably nowhere close to that.
It's easy to beat AWS/Azure if your workloads aren't bursty, I ran a datacenter with ~400 VMs and about 200TB of data (with 50% per year compound data growth) and our cost per VM-hour was way below what it would cost when you included the cost of servers, electricity, storage, networking ports, datacenter buildout, and 1 FTE (we had 3 guys who spent ~10-15% of our time on i
Re: (Score:2)
>So no, it doesn't take massive scale to play with the big boys, most S&P 500 companies could do it.
We perhaps have different concepts of "massive"...
Re: (Score:2)
My employer was on the S&P500 for a time, 650 employees and $7B in valuation, not exactly huge when you're comparing to someone like BoA.
Terrible summary (Score:5, Informative)
Here is a linke to the Q3 conference call on the Motley Fool site
https://www.fool.com/earnings/... [fool.com]
and the relevant bit
---------------
"The cloud journey for Bank of America is an interesting one. We started about really the new BAC framework for those of you, remember that, that came out of that into early days as simplified, improved. We had 200,000 plus servers those sort of accounts now down to 70,000. The first decision we made was to actually create an internal cloud. Those servers operating about 30%, 67 data centers, very much dedicated by line of business, by operational unit, by risk or whatever.
We took all that away and build common architecture, so the lion's share of applications run around 8,000 servers. We still have 70,000 servers, but those are more dedicated for very specific things and we'll continue to work to take them down, we're down to 23 data centers. Those who've been around the Company for, I don't know, we took a $350 million charge a few years in the second quarter of '17 to pay for part of this changeover. But in that time frame, we've reduced expenses by basically around 40%, for $2 billion a year on our backbone. And so at the same time if you looked on pages 14 and 19 you can see just over the last couple of years a volume of transactions. So we're up 86% in mobile logins, we're up 39% wire transactions and things like that.
So what you've had is this scale effect that we've been able to internalize. In our provision these services from what we can get, and the external clouds is still 25%, 30% cheaper, which we expect to change honestly. And so we are working with potential providers to take the next step, which was discrete data centers and resources to internal cloud, save a ton of money, then use that power to actually negotiate with third parties to how they might help you and support you and that's going on with Cathy Bessant and her team right now and how both will run for us.
But so far we're still cheaper and so far we have to make sure that the external providers are safe sound lead the data just for us to use, our customers don't make -- people's data, etc.. And that discussion negotiation goes on, as we speak, but we will -- we're not -- we don't need to own the hardware, we just need to find out who can provide the right way."
---------------
The $2 Billion saved is the old "200,000 plus servers ...now down to 70,000", not in comparison with the cloud.
and "the cloud" is still cheaper and they are looking to go that way "the external clouds is still 25%, 30% cheaper, which we expect to change honestly. And so we are working with potential providers to take the next step,"
Economics at work (Score:2)
Re: (Score:2)
There's this collective insanity about "cloud" solutions. by so many businesses. I work with mid-sized companies and there's this rush to move stuff to the cloud and almost never do I see a hard-headed run-the-numbers approach.
I think some of it is driven by consultants who fear that "the cloud" will steal all their business and figure if they can move shit to the cloud and get a piece of managing it after that they will not lose out.
Some of it is driven by chronically under-capitalized organizations that
Re: (Score:2)
I suspect a lot of it is a result of the chronic sickness in technology of chasing taillights. It's especially common in frontend JS right now, just look at the bazillion frameworks that are the new hotness for about 3 minutes each before everyone collectively forgets about them and moves on to the next new hotness. A lot of frontend development is done by younger developers and a lot of it is "throwaway" work, in that nobody is ever going to maintain it and they'll simply start from scratch in a year or tw
Re: (Score:2)
Finally we'll be on the current stack and free of our legacy bullshit that's been holding us back. Utopia is here!
Re: (Score:2)
>
Some of it is driven by chronically under-capitalized IT departments and think the cloud is some silver bullet that saves them from paying for bad decisions by other business units
I look at cloud and think...we implement the system you want in the cloud, and you just pay for your usage. If your costs are too high, perhaps you need to redesign your system, or reduce your usage. Either way, IT isn't left with the bill of running your crappy design.
Re: (Score:2)
It's sometimes hard to separate other business unit unhappiness with IT from undercapialization.
Business unit A demands some expensive service, but financial leaders never provide the funding to deliver it, and the results are unsatisfying to the business unit, perpetuating the bad opinion of IT and further reducing their access to capital.
My question, though, is if business units weren't paying chargebacks for IT services before the cloud, what leaves them holding the bag for cloud services? It's just ano
Will someone please explain the cloud concept (Score:2)
To these ignorant asshats. They have their own data center. If its the cloud its you using someone else's computer. Occasional it will rains your data on everyone.
"Cloud" does not have to do with who owns what (Score:3)
Cloud computing is about abstracting the hardware away via APIs. Resources (servers, volumes, load balancers) are created via API call. It doesn't matter who owns the servers. If that mattered, then what would you call Amazon.com running on AWS? AWS is still a cloud computing platform, even if the same company owns the e-commerce site and the hardware.
Virtual servers. (Score:2)
Just like we did since the first vserver solutions.
And "it doesn't matter who owns the servers" is such an amazingly insanely stupid statement.
It absolutely matters! It's *essential* to keeping your data safe and staying in control over it.
Your system literally permits a disgruntled third world data center admin to ruin your entire multi-billion-dollar business! That is CRAZY.
Re: (Score:2)
Obviously when he said 'it doesn't matter' he was referring to the definition of 'cloud computing'. And he is absolutely right. A cloud can be public, private, or hybrid.
Re: (Score:2)
Need more power? The cloud is ready.
Another nation to support more? The cloud network is totally ready in that nation..
vs getting permits to build, removing heat, adding cooling, power, moving staff around, buying new computers, networking, adding more staff..
No surprise here. (Score:2)
For a bank with many transactions and widespread operations and specific requirements building an own distributed computer infrastructure is pretty much a no-brainer. Cloud as a service is the first SaaS to make sense, but mostly for people who can only start cheap and are not sure about how far and fast their service will take off. For a Megacorp like a bank this rarely makes sense.
"cloud" has no meaning, ... yet it still lost it m (Score:2)
So BoA is building its own server space? Just like everyone used to do, and everyone sane still does? But wait! No! "It's a cloud!"! ... --.--
I wish I was in the Idiocracy movie. At least they thought being smart was cool, and didn't run behind retards, given the choice. At least there I could make a positive difference!
Re: (Score:3)
What is with these morons who can't read. No, they are not 'building their own sever space'. That HAD their own server spaces (60 of them) with their own servers (200,000 of them). Then they switched to using 'cloud' techniques (shared servers and storage with the servers running VMs) and now they have a lot LESS of their own server space and servers. Which saves money. Get it?
Ineffective management causes high costs (Score:3)
Most Enterprise IT managers are relatively clueless and impotent. They are political animals not business leaders and will take the low risk path. The key term here is "most".
A medium to large company should be able to provide the majority of its infrastructure cheaper than the cloud. The problem is that most legacy medium and large IT shops still expect to do infrastructure the same way with the same vendors and somehow reap the savings while the cloud providers take a different path.
Most cloud providers buy commodity whitebox switches, they buy commodity whitebox servers and commodity disk for storage. They run free hypervisors on this hardware and their storage is exported from commodity servers.
Most enterprises still buy switches from Cisco, servers from Dell and HPE and storage from the likes of EMC and Hitachi. They still run VMware as the virtualisation layer etc. There is at least a 100% markup going down this path so cloud provider has a cost basis which is half that of the enterprise.
Smart Enterprises are copying the models of the cloud providers and unsurprisingly they're finding that their costs can be significantly lower at scale.
The other difficult part is the fact that most large ICT departments are expensive inefficient cost centers dogged by politics. This is true of most business sections however ICT has some real issues. ICT is driven by fads which don't improve productivity, most of the management practices bear more in common with communist Russia than a transparent market. Management are unwilling to address this because it potentially shows them in a poor light.
This is not to say that some workloads such as peak loads are much more effectively deal with in the cloud, however there is the possibility for this to change as well.
The end result of this trend should be interpreted as follows. Don't invest in any traditional hardware and software vendors, they are being gutted and only being supported by legacy players. Invest in the makers of systems or component makers who support both value chains whose market cap isn't dependant on traditional enterprises.
Any business tool/package that is older than 16 years should be critically reviewed with an eye to replacing it with a lower cost competitor.
$2B seems high (Score:2)
Renting 70K servers from AWS doesn't cost $2B.
Assuming they are all midrange m5.4xl's (16 vCPU, 64GB RAM) with 1TB of SSD EBS disk (some will be bigger, some smaller). that's $0.77/hour ($554/month) for the instance + $100/month for storage, or $654/month, $7853 for one year, or $550M/year
Granted, they'll have more costs for bandwidth, other services, etc, but even if costs were double, that's $1B
So, all in they are paying Amazon $1B/year for that 70,000 server footprint, so how are they saving $2B? Even i
Re: (Score:2)
They absolutely can do it fro 1/3 the cost.
" all in they are paying Amazon $1B/year for that 70,000 server footprint"
"Even if they had 200K AWS servers, that' $1.5B total for Amazon and again double it to $3B,"
Well, there's you 2 billion.
Re: (Score:2)
You have to pay for data transfer as well as storage, and you have to pay for fucking DNS routing requests, and you have to pay for extra services that make your shit "elastic" and responsive to demand. Instances are priced differently in different regions. Your segmentation, fault tolerance, data locality (for legal purposes) etc. all impact pricing. You pay more if you're elastic as well, since you're not getting the reserved instance discounts. Or you can fucking reserve and then sell off excess capa
Re: (Score:2)
They aren't paying Amazon a dime, where did you get that from? They used to use the traditional new service=new server method, which resulted in BoA having 200K servers across 60 data centers. They switched to private cloud (not AWS), and now have only 23 data centers hosting 70K servers. So they closed 37 data centers, got rid of 130K servers and all of the cost that goes with them, and saved $2B. At no point, past or present, was AWS involved.
Re: (Score:2)
Double? AWS makes its money on bandwidth. AWS is very cheap, until you start including bandwidth and transaction costs. Every read, write, in/out transfer, datacenter-to-datacenter transfer costs money.
A lot of companies are making those mistakes and not seeing the true costs of Amazon. Unless you can negotiate a deal with them, AWS is insanely expensive.
In house development (Score:2)
Is almost always cheaper then 3rd party in the long run.
Why cloud is cheaper, and why AWS makes so much (Score:3)
AWS is hugely profitable, and most companies save money using AWS instead of running their own computers.
The reason really has to do with scale. Everyone knows that scale matters for the physical aspects (power and machines and buildings), but it matters for people possibly even more so.
If you're willing to build and staff a data center in the middle of nowhere it can probably be close to as efficient as Amazon's data center's - by "close" I mean well under 2x, not 5% or something. This is based on mghpcc.org, built by the 5 big Massachusetts universities - 15MW, $100M for the building, competitive power costs, room for 20-30K machines. Based on the public numbers, amortized over 20 years, the building costs less than $200 per 1U per year. At $0.10/kWhr (public tariff for MW rate * 1.2 PUE) power costs about $1/W/yr; our servers seem to draw in the 300W range or less. Since servers seem to last 3-5 years, that means the machines themselves cost considerably more than the building or the power, but not a huge amount more.
The MGHPCC costs about $5M/yr in depreciation; you could probably make the building smaller and cheaper, but the cost per computer would undoubtably go up, and there's not a huge amount of room before it starts adding a lot of per-computer cost - my wild-ass guess is that below half this price it would start getting really expensive. (I'd really need an industrial-scale A/C expert to make that prediction)
The other place where scale comes into account is your workforce, because people are expensive and they come in integer units. If you want to use serverless, and elastic mapreduce, and block and s3-like storage, and virtual machines, and a couple of database solutions, and a load balancer, and a few other of the services that are trivial to select on your AWS dashboard, then you are going to have to hire somewhere between a dozen and a few dozen highly-qualified people to set them up and run them. (among other reasons, because unless you're Google or Amazon the very few people who can do all of these won't want to work for you) At this point you're talking a payroll of millions of dollars per year.
Add the payroll to the depreciation costs of the building, and you've got your fixed costs; to achieve vaguely Amazon-like efficiency this probably needs to be 25% of your total costs or less. Working it all out, if you've got more than maybe $50 million worth of computers (preferably lots more), and can hire good people, you can run an operation close to as efficiently as Amazon and save lots of money. I would think Bank of America falls in that category. In contrast, for non-huge companies Amazon is a bargain, because you're in effect buying a small fraction of not only a large data center, but of each of the folks who maintain the services you use.
Re: (Score:2)
Even with a few IT persons, you can rent space in datacenters and run a pretty good operation, either dedicated or colocated servers will still be cheaper than whatever Amazon has to offer.
AWS makes sense for quickly scaling up, load balancing, as disaster recovery and testing. But you can do that just as well on DigitalOcean for a bunch cheaper.
The true meaning of cloud (Score:2)
I'm kidding. The term is completely meaningless.
Those who wish to effectively communicate with peers avoid using meaningless words. Avoidance benefits everyone by minimizing unnecessary confusion and ambiguity.
Two examples illustrating this point:
Piss poor: "That shit over there was all moved to Azure cloud"
Better: "All of Rack E was moved to Azure IaaS"
Notice word "cloud" is just as worthless as "shit" in the "piss poor" version. "Better" not only provides more information its actually shorter.
Piss poor
I Believe It (Score:2)
The cloud is great until you look at the price tag (which is a surprise at the end of the month).
Go ahead, try using Amazon's "simple" calculator. https://calculator.s3.amazonaw... [amazonaws.com]
Don't forget to include ALL Amazon cloud services you need (see the 29 items on the far left side of the page).
What's that? You don't know what these things are, what they do exactly, or what you need?
JUST AS BEZOS PLANNED.
Hell, even if you spend the 2+ years to get certified 6 ways from Sunday and figure it all out, by the time
and in three years??? (Score:2)
Next story about BoA (Score:2)
Security (Score:3)
Re: (Score:2)
The newest definition of a private cloud just means that the resources are logically separated from the other parts of the cloud. Personally I think a private cloud means that it is your servers running on your network, but the big cloud providers disagree, because to them only their clouds are real. Why would you want to run your own cloud when you can pay Amazon/Microsoft 5 times more to do it for you?
Re: (Score:2)
I think this is a really fair question, and there are some differing opinions abut what it really means.
Personally, I interpret "private cloud" as any time you host your own server that allows not just yourself, but OTHERS to do the things they're typically paying businesses to do for them with traditional cloud hosting.
For example, I run a copy of NextCloud, a free open-source product that essentially mimics the file storage and sharing capabilities of a service like DropBox, plus provides extensions to sy
Re: (Score:2)
Pretty much, and the usage is almost as old as "the cloud" itself. It's a marketing term that never actually meant much more than "on the internet" in the first place, and its only devolved from there.
Re: (Score:2)
"Cloud" is a marketing buzzword. What it means is whatever the marketeer thinks you want to hear. That's sadly the reality, but let's put the real-world cynicism aside a moment.
Many companies are using their own resources to basically duplicate the commodity infrastructure and software ecosystem that Amazon, Microsoft, &etc. sell, and calling it their "private cloud". The nomenclature keeps the PHB types from having an impotence crisis, lol.
For programmers and end users it is functionally identical t
Re: (Score:2)
"Calling your own servers a "private cloud" is kind of like calling your car a "non-car." "
Bad metaphor. It is like calling your car a "private taxi".
Re: (Score:2)