Gartner: Stop worrying and love the cloud, with all its outages and lock-in

Stop worrying about the resilience of clouds, the possibility you'll be locked-in, or the systemic risks they pose – and instead learn to appreciate their benefits and limitations. So said Alan Waite, a managing veep of analyst firm Gartner's Technical Professionals advisory service, at the org's IT Infrastructure, Operations …

[martinusher]

Bad analogy

The statement about airlines presupposes that the person making it hasn't tried to travel anywhere using a US domestic carrier recently. They'd find it expensive and frustrating, starting with schedules that are likely to be changed without notice, routes that turn a fairly short journey into a all day ordeal, lackluster service and everything maxed out (and with a nagging feeling that the quality of both planes and crews is starting to leave a lot to be desired).

This might be stretching analogies a bit far but as it turns out the reasons for the problems for airline travelers feeling perpetually screwed are not unlike the same reasons cloud customers might harbor the same feelings. In both cases ruthless consolidation has led to a paucity of suppliers who might appear to be competing with each other on the surface but in practice have the market carved up. Their primary task is to monetize everything they can, squeezing customer yields by giving the minimum and charging the maximum. Customers don't really have a choice, its really down to 'travel or 'not travel' with driving more an act of desperation. Customers for cloud services may find themselves in the same boat -- clouds were only cost effective while corporations were building market positions; once solidified then there's no incentive to do anything except charge as much as possible for as little as possible.

[headless_chicken_omg]

Re: Bad analogy

Aweful Analogy... which vendor is sponsoring this?

t's a far easier thing (to a degree) to replace your hardware vendor (i.e Cisco) and / or VMWare (yeah ok little harder) than it is to move your DC, Cloud Provider etc. Particularly if you have cloud-native code thats in play!

[Anonymous Coward]

Re: Bad analogy

Moving into a cloud is a year long project, full time to anyone with significant amount of software/hardware running.

Replacing Huawei or Cisco switches from corporate networks? Less than a month: That would be trivial compared to moving out from a cloud, any cloud.

[John Hawkins]

Of course he would say that

But in a year or two when it's time to renegotiate your Cloud provider contracts, would you rather be locked in or be able to migrate your infrastructure to a rival provider at a reasonable cost?

Given the amount of money one provider is throwing at a project I'm currently working with to 'enable' migration, marketing and sales have very deep pockets; that money has to come from somewhere and I'm guessing that locked in customers are at least part of the answer.

A lot can be achieved without having to use provider specific solutions - Kubernetes anyone? - and you're probably going to end up with a simpler solution as well.

[Anonymous Coward]

Re: Of course he would say that

I was working for a company that moved whole thing from machine room operator to another machine room operator (and saved few hundred grands per month). Well worth doing. Took only few months: Virtual machine images are, you know, portable, so there was only 20 or so major servers to be moved. Moving databases took the longest time.

Trying to move anything now working in cloud x to cloud y is at least 10* more work. And cost. Vendor lock in is absolute and perpetual.

[Anonymous Coward]

Re: Of course he would say that

Yeah, the authors own admission is that their advice is for operations that "could create a lightweight version of their website based on static content" to consider his approach.

So like 1 in 10,000 use cases or less across the cloud hosting industry over the next 24 months? Assuming that the companies are either not using any off the shelf software of can afford to write a "lightweight" version of third party software (if it's even legal).

I mean, sure, nice work if you can get it, but re-architecting an entire enterprise, cakewalk that, no risks of downtime at all right? And I don't buy the numbers that we are seeing a long term downward trend in cloud reliability. Those systems have reached a level of byzantine complexity that defies analysis at the scale of AWS, Google, or Azure. The cloud providers in China have layers of market and political uncertainty as well.

As always, beware taking candy or advice on enterprise software and hosting services from strange analysts.

[Timop]

How convenient to pick aviation as reference.

Aviation industry has common goal of high security standards and extremely low accident rate. This increases viability of every aviation company.

What if they could have major vendor locks with monthly billing no matter if someone dares to fly or not and for those who must it does not matter how bad accidents will happen because migration to competitor would still be worse option financially?

[computing]

Good point.

Also, safety depends on how you measure it - per-journey, or per-mile.

When measured per-mile, aircraft win of course.

But when measured per-journey, cars are three times safer than air travel. So if you flew to work each morning (think air-taxis), you'd be thrice as likely to never come home than if you drive in.

Source: https://www.telegraph.co.uk/travel/rail-journeys/trains-are-better-than-planes/

[b0llchit]

So much positivity... This is a case for Follow the Money analysis.

[Anonymous Coward]

What you mean is "How much Microsoft paid for this piece of advertisment?".

Because that's what it is.

[dubious]

multi-cloud

Was that essentially "Don't do multi-cloud and go all in on your one cloud pusher 'cause they're reliable, except actually maybe go ahead and build an entire separate system on a different cloud just in case your primary cloud fails."

[Anonymous Coward]

why does ANYONE listen to these morons?!

More importantly are Amazon, Google or Microsoft clients of Gartner and "encouraging" them to make statements like this?!

Cloud is massively expensive, a huge security risk and once you're in, you're screwed.

This is the kind of comment you'd expect from a suit whose probably never worked infrastructure.

If you have a DC running for example Dell Hardware with VMWare and Cisco networking it isn't THAT hard to replace it with HPE, MS Hyper-V, Juniper for example using internal staff during a normal refresh cycle.

Try pulling your data out of a SaaS database and rewriting your entire application stack to do it.

I hate these consultancies. What is the genuine bets that within a few years Gartner will be advising clients to move away from the cloud to inhouse or datacentre hosted servers after a couple of good well publicised outages/hacks of Cloud vendors?

I work somewhere where I had to sit through a "consultation" with one of these idiots and apart from "You should be putting everything in the Cloud" there was the usual "your teams can then innovate" you should be listening to us so you can then go ahead and innovate.

[xyz]

Re: why does ANYONE listen to these morons?!

Wot he said. There is jack reason to stick everything in someone's "cloud" esp if you've got stuff across multiple locations that are a mix of on-prem, hosted and cloud. And as for Gartner, last I read from that font of wisdom (last week) was cloud is "old", so... what to make of it all. ;-)

I actually love the disclaimers at the end of any Gartner presentation... In effect, we didn't say anything, we weren't here, nothing to do with us.

[JohnSheeran]

Re: why does ANYONE listen to these morons?!

Gartner is a completely worthless source of information. You have to remember that they get a part of their funding from the people that pay for the service from them, but they get the rest from the people that pay them to write these articles.

My company has recently undergone a "benchmark" exercise with them, and they didn't even know how to benchmark our cloud usage.

Over the years I've always heard from friends and acquaintances that work for large tech companies that they pay Gartner to say what Gartner says.

[sgp]

Re: why does ANYONE listen to these morons?!

That's the magic part of the magic quadrant. All rankings based on payments.

[Nick Ryan]

Re: why does ANYONE listen to these morons?!

For many years I've applied the truism that the only useful thing about any Gartner report is that they waste a bit of time while being the first to guess who paid for the report.

[Potemkine!]

Another bad analogy

"Perhaps it was VMware in a datacenter. Perhaps it was Cisco for your network. You were happy to accept that lock-in because those vendors gave you fantastic capabilities that you could leverage and use."

Most of the time, you don't rent these equipments, you buy them, once for all. You know the price at the beginning and the investment is well defined. With the cloud, price can change from one day to another with no easy way to move to the competition to reduce the cost.

And frankly, moving from VMWare to Hyper-V or from Cisco to HPE isn't that hard.

[katrinab]

Re: Another bad analogy

Or even if you do "rent" them, it is only rent for tax purposes, in reality you have a multi-year payment plan with a bank and you effectively own the thing.

And the decision as to whether to "rent-for-tax-purposes", that depends on the tax rules at the time of purchase, and they change frequently.

[Doctor Syntax]

Re: Another bad analogy

"you effectively own the thing"

Even to the point of having to work out how to dispose of the remains at end of lease.

[43300]

Re: Another bad analogy

Indeed. If you wanr to replace switches and routers - well, you just do it. And last time I looked there were tools to convert VMs between the common hypervisors (although I'm not sure that many people would want to move from VMWare to Hyper-V - I'm familiar with both of them and would always choose VMWare).

[SammyB]

Can I borrow your watch

These kind consultancy statements always reminds me of the old joke about the consultant asking you to borrow your watch and then telling you the time.

[Doctor Syntax]

Re: Can I borrow your watch

There's a good reason for that type of consultancy.

High salaried manglement want to know what's going on in the business and how to improve it.

They could look at the watch - i.e. ask the workers ho know.

But workers aren't paid very much

And the high salaried manglements aren't going to encourage the view that the value of anything is might not be the same as its price.

What to do?

Simple, commission a consultant who can borrow the watch - i.e. talk to the workers - write up what they said and present it as a reassuringly expensive report.

Problem solved.

[Duncan10101]

I'm probably my own worst enemy but ...

I kinda agree with him. I know, I'm about to get a deluge of down-votes, but the thing is, being from an annoying consultancy who makes bad analogies doesn't actually make him wrong. If you look at cloud "Outages" over the years, most of the down-time is caused by all customers going against the cloud-vendor's advice, and putting all their eggs in one basket/DC. His advice that you should use the existing technology to mitigate the risk of downtime is actually good, and more people should take heed. Distribute your servers, and the chances of failure are amazingly low. So the real problems are lock-in and expense. I also agree with him that lock-in isn't such a biggie, because you'll get some form of that no matter what you do. So ... expense. That's a personal choice, I won't judge.

[Roj Blake]

Re: Distribute your servers

Distribute your servers and the chance of a failure actually goes up. It's the impact of a failure that goes down.

[Duncan10101]

Re: Distribute your servers

Seriously? I think it's quite obvious that when I wrote "Failure" I was referring to the overall system, not a single, redundant component. So the be clear: I'm advocating redundancy across multiple DCs ... and that is exactly what the cloud providers suggest. Example: search for AWS failover regions.

[Tim 11]

Re: I'm probably my own worst enemy but ...

I've got sympathy for your views - I think cloud definitely has many advantages in some situations but IMO lock-in is on rather a different scale

If you've built your app as on-premise .NET say, you may be locked into Microsoft but you're not locked in to a server or hosting vendor, and assuming your servers are virtualized you have lots of options for deploying them, in the cloud or on-prem, without changing a single line of code. However if you've built it on AWS and want to move to azure, you're looking at a significant rewrite

[Claptrap314]

Re: I'm probably my own worst enemy but ...

You're also looking at a couple of orders of magnitude jump in outages, if that matters...

[Anonymous Coward]

Re: I'm probably my own worst enemy but ...

>> If you look at cloud "Outages" over the years, most of the down-time is caused by all customers going against the cloud-vendor's advice,

That's nonsense. While customers tend to cause most cloud-related security incidents, they normally don't cause outages (and, frankly, if your customers are able to bring down your cloud infrastructure then that shows your resilience is non-existent and you shouldn't be in business, period!).

Most cloud outages are actually caused by infrastructure breakdowns (Amazon, GCP) or simply configuration errors/stupidity/ignorance of the cloud vendor's IT staff (mostly Microsoft). Azure (which really is the worst of all cloud vendors) sees regular outages caused by misconfiguration or deployment of bug-ridden software pieces, resulting Azure having lower uptimes than any half-decently managed on-premises installation. Just ask MS365 hostages how often it has been down this year.

If you move into the cloud expecting better uptimes then you're fooling yourself. What the cloud gives you is the *idea* (empty promise) of resiliency while, in reality, you're paying an increasingly hefty ransom for a vendor who is contractually obliged to not give a fuck when things go haywire (and they do quite often).

Nothing of this is new, though, and at the end of the day cloud providers are just businesses who want to lock you into their offerings (make it easy to join and difficult to leave) while extorting an increasing amount of money from you. There certainly are situations where cloud deployments make sense, but they are the exception rather than the norm.

[Claptrap314]

Re: I'm probably my own worst enemy but ...

You are right about Azure, but wrong about resilience.

AWS & GCP give you the tools to get resilience. But you will never get to four nines (let along five) in a single region. AND THIS IS WHAT THEY TELL YOU IF YOU BOTHER TO READ

Of course, they also do barking-insane "serverless" and "noops". That's marketing, and I really wish they would stop.

If you want HA (4+ nines), you are going to have to architect your entire stack for that. You cannot take an existing app that runs on one server, port it, add autoscaling, and expect it to be HA.

The fact that to be HA means that you need 24/7/365 pager coverage does not change. But if the bean counters think otherwise, well...

[MTimC]

Waite is, substantially, correct. Few orgs measure their on-prem RTO/RPO very well, nor allocate costs appropriately. I've only encountered one client that had a DR process that was thoroughly tested before a disaster, and that involved doubling capex, by swapping prod and DR each month. Cloud outages get huge publicity.

I'm not sure about 'lightweight versions of services' this sounds expensive and challenging to get sufficiently right. It's usually easier to focus on getting blue/green deployments working well and use this approach to underpin failover to a different set of datacentres.

Lock-in is always a problem and needs someone to worry about it/measure it. For clouds, it's usually better to take advantage of new services, and then backfill them with alternatives post implementation.

[Doctor Syntax]

I'm sure there should be scope for a website that tracks these consultancies - what they say now vs what they said a week, a month, a year and more back, how consistent they are and how their predictions turned out.

[Anonymous Coward]

The lack of rigorous history / tracking is how some of these so-called experts and consultancies stay in business.

They also rely on the fact that greybeards are a shrinking population, and even though it cycles in replacements from time to time, the people with organizational and industry tribal knowledge (and scars from hard-won experience) are often not C-suite decision-makers, usually tech leads at most.

Meaning that the likes of Gartner, McKinsey et al are often able to publish reports or strike deals by waving new shiny things around in the executive offices (or golf courses) without scrutiny from grumpy old veterans who remember the last time some windmill tilt from the consultants fell down.

[Paul Hovnanian]

Lock in?

"If you've if you've worked in IT for any period of time, you've always been locked in in some way"

Not so much. Develop in a heterogeneous environment to some architecture agnostic standards and you'll be OK. Apps developed on a Linux desktop and moved to production HP-UX, Sun and AIX servers. Not Linux at the time because management still had their panties in a bunch over open source. Just don't tell them that the web stuff was Apache and Perl running atop their precious commercial systems.

Lock in? Hardly, as most people outside our team had little or no idea what brand of system they were talking to.

[Anonymous Coward]

"People accept lock-in understanding the risks that are involved with it. "

This statement alone tells us that this guy is a paid mouthpiece of a cloud provider. Most probably Azure as Gartner and Microsoft have and have had 'good relations' for decades: Microsoft pays and Gartner tells us what they are paid to tell.

Very simple operation and win-win for G and MS.

[CowHorseFrog]

Heres a brilliant idea, lets take advise from a company that has never written a line of code and yet they claim they know everything about software. They know even more about products they are paid to recommend.