Puff on a hybrid – next thing you know, you're hooked on a public cloud I had a flash of inspiration today. Hybrid public-private cloud systems are becoming a gateway drug to pure public clouds. Why is this an arguable view? Let’s look to the ideas discussed within The Big Switch, written by Nicholas Carr. In his seminal book, Carr argues that a public IT utility provider will provide CPU cycles …
Although movement towards public clouds may seem inevitable, companies that work with a hybrid, community or private cloud scenario are certainly not excluded from doing public cloud later.
If anything, a company's previous experience with private cloud solutions will make them very careful shoppers. Public cloud providers may need to work harder to win their business. And companies who choose private cloud technology can also work out their initial growing pains on their own terms and on their own-premises.
CAPEX and OPEX price tags are not the sole considerations for companies already running their own data centers., security, confidentiality, control and many other factors will also play a role in the business decision.
Whereas, for a startup, the decision to use cloud services is usually a no brainer.
Does a public cloud have a solution to the issue of security? A companies secrets are protected to an obvious degree when using in-house servers.
Software licenses. Are they aware of the cloud. For example If I could purchase all-you-can-eat licenses for an office or a site, will it have to be re-negotiated for use on the cloud?
It all boils down to trust. The biggest issue of trust for a company is the level of trust it puts in its employees. Far and away the greatest cause of data security breaches is accidental or deliberate action by an employee.
One example is the use of unencrypted emails as the file sharing mechanism of choice both within a company and with the company's customers, suppliers and partners. It doesn't really matter where your email service is or how tightly its security is locked down. Sensitve data is routinely launched in the clear, into the wild.
Now add BYOD smartphones and tablets and things get even more interesting.
A company with substantial in house systems puts a particularly high level trust in its IT administrators, who can, and regularly do, make mistakes, or worse, deliberately sabbotage systems or discolse sensitive data. Hell hath no fury like a sysadmin sacked.
Where a company's systems physically reside, other than the obvious data protection constraints on geophraphic location, is in many ways the least of its data security worries. In this respect, the use of an enterprise class cloud service is at least as good as using your own data centre.
I've never been convinced by the Utility analogy. With electricity, gas or water, a resource is piped into your premises, and then you run the process yourself. For example, using water and electricity in your own washing machine to wash your clothes.
With cloud, it's you sending the resource (data) to the cloud service provider, the processing is done on their premises and the results come back to you.
So (and I do realise this doesn't meet the needs of cloud vendors and analysts at all), a better analogy would be a laundry.
We just ran a cost analysis of the price to migrate to AWS, and it ran to significantly more than our annual cost to support our environment, plus we actually own our servers, storage, networking equipment, etc., so we get long-term value out of that capital investment as opposed to paying year-on-year for a service which can be turned off at any time. Not to say that cloud services can't be valuable, but sometimes it's worthwhile to make the investment in capital and expertise.
This post has been deleted by its author
"I had a flash of inspiration today. Hybrid public-private cloud systems are becoming a gateway drug to pure public clouds. Why is this an arguable view?"
Because you miss a few very important things, Chris.
1) The cost of bandwidth. Bandwidth costs much, and - shock of shocks - there are plenty of reasons that companies need to send large quantities of data to manufacturing sites, knowledge worker sites filled with video/photo editing staff and more. The cost of bandwidth isn't coming down for the plebians any time soon, so cloud computing is still "expensive resources on the end of an even more expensive resource.
2) Economic Espionage. NSA, GCHQ et al. I'll leave you to work that one out yourself.
3) Trust. Amazon, Dropbox, and Dropbox again. Microsoft, Microsoft again, and again, and again, and again. Salesforce, and Western Digital and on and on and on. But most of all - above all other examples - Nirvanix.
4) Latency. Tier one apps doing active-active where the speed of light is holding up transactions? Me gusta.
5) Disaster recovery time. Remember that part where folks actually do require onsite data, no matter how much the cloudy providers hope and wish and wail and gnash? Ever tried to suck 1TB down an ADSL connection? How about 10? 15? 100? If ($time_to_recovery > $time_to_customer_loss_during_outage) { run $you_are_fucked; }
6) Cost. "Cheaper than owning and supporting your own kit?" [Hearty belly laugh].
7) Pay or die. Economic downturn = can't pay subscription = "you're fucked". On premises = "you can sweat your assets." You might as well say "renting is such a great idea that nobody in their right mind would buy their own house". I suspect different people have different priorities. Let's talk to San Franciscans about the variability of rent over time, hmmm? All markets collapse into an oligopoly over time. I'll be handing my testicles to Amazon or Microsoft on a garnished platter, thankyouverymuch.
I could go on. And on and on and on and...
Look. The future is emphatically, absolutely, and without question not going to be a pure "public cloud" world. Hybrid? Yes. No technology since the introduction of the mainframe has totally replaced it's predecessors. Supplemented? Yes. Supplanted? No.
And get off my goddamned lawn!
I agree with many of these points. Because of latency and bandwidth, all pieces would reside
in the datacenter you are accessing remotely. Trying to do the hybrid thing is very expensive.
Regarding DR, I would hope there would be next gen solutions for that SRM on steroids or something
similar.
Okay, I need to print 4TB of high-resolution photographs per day at my photolab on printers the size of cars. Please explain how I am going to run all my servers in the cloud and stream the images I need back to my photolab. Consider that the lab in question probably clears about $5M a year and wobbles on the edge of profitability as is. It can't afford a bigger pipe than it has now. What magic do you use to make this work?
Now, let's look at my machinist shop which has the same sort of requirements; data that must be delivered to local equipment in a timely manner from the cloud. The next-generation stuff does a closed-loop between the manufacturing equipment, sensors and analysis software which needs ultra-low-latency in order function properly. Am I going to run all of that in the cloud?
I also have bakery that falls into a similar category. These folks do a million samples a second from their sensors across the whole of the factory then crunch that data in real-time and feed the results back to the machinery for real-time modifications of the environment. Are you going to do that all in the cloud?
I have remote drilling teams that are doing real-time seismological analysis, modelling and simulation based on feedback they receive from on-site sensors. This information helps them decide where to drill, how and when. Their access to the internet is via an orbiting dirigible with an LTE booster. Are you going to put their workloads all in the cloud?
I have a storm chaser that collects over 50 billion samples a minute from over 1000 sensors and crunches that in real time to determine how storms are going to evolve. He is often driving between mountains where even cellular signals won't reach and satellite is thready to the fact that he drives into tornadoes for a living. Are you going to put his workload into the cloud?
I have a journalist that deals with Chinese dissidents, pursues human rights violations by the American government and is currently trying to uncover some unspeakable horror in Burma. Even if you could put all his workloads in the cloud, would you?
I have a fire hall that absolutely has to have the diagnostics and maintenance systems for their equipment running 60/60/24/7/356, no exceptions. They need 100% uptime and access to a number of emergency systems and are increasingly using sensors ranging from deepscan sonar to thermal sense drones to determine safety. Are you going to put their workloads in the cloud?
I could go on and on and on, but suffice it to say you're talking utter fucking bullshit. Some workloads can be put into the cloud because they have no localized mission criticality. Some workloads absolutely can not. Even for SMBs - like my 10 man bakery - there are workloads that will run local and some that could be moved to the cloud.
But the cloud is a tricksy thing. If I have workloads that I must run local - and despite your propaganda this will always be a truth of the world - then I have a floor cost of investment in local IT that I must make. If I am already balls-in on some local IT, then the question becomes "do I have the spare capacity on my local setup to run $_workload or not?"
If I have the spare capacity to run $_workload locally I do. Period. It will be cheaper to do so than farming it out to the cloud. If I don't have the spare capacity to run it locally then I ask myself the next question "what is the cost of running this locally versus the cost of running it in the cloud?" I already have local systems, local nerd and the rest...if the TCO of adding that workload locally is lower than farming it out, it gets added locally.
The cloud is great for DR. That way I don't need to light up a DR site. Provided, of course, that everything is encrypted at rest as well as in flight, and that data sovereignty issues are dealt with. And that I can download the data to my local network - where it will inevitably reside once I light my factory back up - in a quick and financially painless manner.
Some workloads that are finicky and irritating, but not especially mission-critical - like email, instant messaging and so forth - I have no problems putting into the cloud. The world doesn't end if e-mail stops for a day or two because Amazon blew up. My company does stop working if the delicate dance of complex sensor-analysis interactions with the bakery machinery ceases.
And if I can't get the fire alarm notice, why then...people die>.
How about you get "off message" for a little while, stop thinking like a marketdroid and start thinking about the human impact of cloud computing. The cost in lost profits from downtime, the cost in jobs from lost profits or shuttered businesses and the cost in lives if some things go wrong.
Then tell me, with a straight face, that the future is to have all workloads in the cloud. Because if you can actually do so you are going on the blacklist of "IT professionals" that I will never, ever deal with...and association whit you wilt be the viral touch of death for any contracts, vendors and so forth that I deal whit in the future.
You have a whole great big box full of tools at your disposal. Don't keep using a hammer for everything because it's what you have in your hand at the moment.
I wouldn't pretend everything belongs in the cloud. Certainly not with some of the workloads and access you describe. For others, they felt they had no choice. Netflix being the poster child.
http://bit.ly/1gVnxqX "The overriding reason for the Netflix move to a public cloud generally, and to AWS in particular, was that the company could not build data centers fast enough to meet the often-spiky demand of its users and that AWS was the only game in town in terms of the scale that Netflix needed." More details about that here and about - Google is your friend.
But yes, there will be corner cases and situations where it will not be a good fit. But the cloud folks are bending the cost curves down and the bean counters will count the numbers and the dash will begin. I'm not saying the dash is underway - but it will happen. "Then tell me, with a straight face, that the future is to have all workloads in the cloud." Maybe the reasons won't be good, but the numbers folks rule the day in the long run there will be a tipping point, I'll bet you on that and it will become quite apparent.
"The cost in lost profits from downtime" Yes. I often use this example of a cable cut and the havoc that resulted: http://bit.ly/1eCCFJT I expect there will be SMBs that totally moved to the cloud, cables are cut and folks have to work from home for a few days. Yes high profile burps in Amazon, but those are becoming less frequent as they tighten their processes.
For quite some time to come "100% in the cloud" will be the corner case, not the norm...and the cloud will never fully replace "owning your own gear" for all businesses.
New technologies are a supplement to exiting ones. They don't supplant them.
And bear in mind it's a lot easier to start a brand new - and entirely virtual - company (such as Netflix's VOD offering) as a 100% cloud-based service. Hell of a lot harder for an established business or one of the many that have on-premises needs for IT.
Will 100% cloud be a possibility for some companies? Sure. Will 100% cloud be a possibility for most companies? I doubt it. Will 100% cloud be cheaper than on-premises for any excepting the odd corner case? I sincerely doubt it.
The public cloud is great for (mostly American) companies that will either not exist for long (such as a political campaign) or those (mostly larger) companies whose internal politics is such that buying a capital asset is a miserable pain in the ass, but paying rent is (for now) under the radar. (Though finance people will catch up to covering that in red tape eventually.)
And again, none of your pro-cloud rah-rah even begins to address the issue of sweating your assets. Can't pay your subscription? Fuck you then, go out of business.
No, I think that even if it were cheaper - and it's not - and even if it were feasible - and for many, again, it's not - lots of businesses would chose to avoid putting 100% of their workloads in the cloud.
Boil it all down to it's purest essence and it comes down to risk aversion. Most people who run businesses do so because they want to be in control of their own lives. You'll find a significant % of them don't want anyone's iron-gripped hands around their testicles, regardless of how "household name" the American super-company paying the marketing dollars in question is. When you are 100% public cloud, then a little squeeze and you beg for mercy.
Though, hey, if you like it rough...
Spot on.
I do think a convergence in software development methodology (with horizontal scaling and resilience in the software rather than the hardware layer) is inevitable and agree with the author that this poses a grave threat to the existing hardware vendors, but the step from there in which public cloud conquers all is badly flawed for all of the reasons you stated.
I do foresee a growing number of build, own and operate requirements for on-premise cloud type infrastructure outsourcing structured in much the same way as classic mainframe contracts for Sun, IBM, HP etc. look but with the difference that the infrastructure is identical to that which could be run externally.
Aside from that the public/private question on cloud is extremely similar to the wider outsourcing discussion. Outsourcing turns a complex internal risk assessment and management question into a complex external risk assessment and management question in which capex vs opex and distance, data security and the relative effectiveness of SLAs versus internal controls.
In the same way that effective outsourcing needs strong and experienced procurement and governance, tends to cost more than it looks like it will and usually only works for non-core business functions I'm sure the same will prove true in the long run for public cloud.
excellent points. Two more:
- increase risk for the global economy. If a traditional IT structure
fails, this can be a disaster but only affects the company. If a cloud
service fails, it can bring down the entire economy.
- increased long term costs. Yes, the prizes go down. We are in the
hook up phase like a drug dealer dumping prizes until the customer is
dependent. Once the local IT structures and cultures have gone, companies
have no choice anymore. And then, the prizes will go up. And there is no
way back.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
