[nemothekid]

A common post I see on HN is "why is Company X using the cloud, don't they know, at their size, doing it themselves would be cheaper?"

As more and more companies are moving to the cloud, I'm assuming that a lot of smart people have done the math and have decided Cloud is more efficient.

[dijit]

The truth is somewhere in the middle.

I’m a sysadmin, I know hardware. I think it’s a complete myth that hardware is hard: especially compared with the irreducible complexity that is AWS.

But: I find myself coming back to the cloud.

Why? It costs more and you have less control. Scale up is not as important as it seems and the 10x cost difference would mean scale is not a factor either.

But, in my experience, not dealing with an IT department is the main reason.

Hardware lead times can be high, sure, but there is nothing more frustrating than depending on someone to do their job and they do it bad.

The “efficiency of scale” is almost entirely in the tooling, certainly not the cost, when it comes to cloud.

I keep coming back because even though compute is not hard, nor networking for that matter: distributed storage is still hard.

I have three on-premise servers now and dealing with the IT team is.. well, let’s just say I prefer to deal with the cloud vendor.

The number of people hired to manage the cloud tends to be the same number as hired to deal with on-prem.

I've been in a company with many hundreds of physical machines to manage and a team of 4, and I've been in a company with infrastructure that handles a similar use-case on cloud in a team of 5.. So I don't buy the headcount argument honestly.

[Alupis]

Great points, but also consider for medium and small (and large) businesses, being able to provision and pay for only what they need today is often cheaper than spec'ing out beefy hardware to handle tomorrow's scaling issues.

After all, most companies using the "cloud" still have on-staff IT anyway, since non-technical people still cannot manage AWS or any other cloud provider on their own.

Cash flow is another thing - lay out big money for redundant on-site servers and supporting hardware today, or pay as you go and use the cloud. Even if the cloud is ultimately more expensive, it's easier for C-suite folks to plan for a small monthly expense vs. huge up front expenses and then dealing with failures and upgrades, etc.

[chii]

> pay for only what they need today is often cheaper than spec'ing out beefy hardware to handle tomorrow's scaling issues....plan for a small monthly expense vs. huge up front expenses

this is probably the biggest "value" in cloud infrastructure. Capital expenditure is difficult to obtain, but debt is both easier to obtain (esp. in such low interest rate environment of today), plus debt is tax deductible immediately, while capital expenditure is at most a depreciation over the expected lifetime of the hardware.

[tw04]

> Great points, but also consider for medium and small (and large) businesses, being able to provision and pay for only what they need today is often cheaper than spec'ing out beefy hardware to handle tomorrow's scaling issues.

Pretty much every major vendor has or is building an as-a-service play so that you don't have to buy today for 3 years from now. Imperfect for sure, but getting there.

[alberth]

Why not then lease dedicated servers from a hosting provider?

You still get ridiculous cost savings over cloud, and none of the long lead times you mention when buying your own hardware.

[olyjohn]

Like, why do you have to do one or the other? Why not use a little of both?

For example, we knew the limit of our big fatty pipe coming into our building. 99% of the time, we could handle all the traffic coming in. The other 1% of the time, we just offloaded the extra traffic to AWS. All the cost savings of on-prem, all the scale benefits of the cloud.

[dilyevsky]

And then when you get large enough and hire dedicated “devops” team to manage your cloud and now you’re back to “dealing with IT department” (rebranded) except now you also pay through the nose for network and storage and anything managed. Oopsies.

[imwillofficial]

It’s obviously not that simple.

Like almost everything it’s a trade off.

I see a lot of underpaid, short staffed IT staff moving to the cloud because it takes massive swaths of headache off their plate.

[dilyevsky]

I think you’re def right that deferring the responsibility is part of it. Now it’s not “IT dept is understaffed and underpaid” it’s “cloud provider’s bug and everyone else is down too so give us a break”. Cloud seems happy to take the blame but they want premium for it. In the end leadership aren’t really saving anything just appear more competent to their bosses.

[dgudkov]

It really depends on what cloud services are used. The strategy of all the major cloud providers is to attract customers with cheap low lock-in commodity resources like S3 or EC2 and convince them to start using high lock-in and high-margin services like serverless computation. If they succeed (and they frequently do), the bills go through the roof.

Read also: https://a16z.com/2021/05/27/cost-of-cloud-paradox-market-cap...

[garethmcc]

I agree somewhat, but its a little more nuanced. Right up front, you could absolutely get infrastructure setup yourself using things like EC2 etc that would eventually be cheaper when traffic gets high enough but there this does not take into account a few things: 1. The cost of paying the professionals to get this all up and running 2. The delay in getting solutions out to customers to generate revenue since the dev team needs to wait for infrastructure to be spun up first. 3. Ability to handle spikes in traffic. Traditional VM and even container based architectures can take minutes to spin up new infrastructure in response to sudden traffic spikes.

And the comment about "the bills go through the roof" aren't 100% true either. AWS isn't sitting around trying con you into spending all your money. If they were they wouldn't have: 1. Reduced the price of AWS Lambda multiple times since launched 2. Switched from billing for AWS Lambda in 100ms increments to per ms increments, saving everyone using a lot of money. 3. Developing an alternative v2 solution to API Gateway that is faster, easier to use and cheaper per request than API Gateway v1 was. 4. Constantly bringing prices down for all services such as S3, SNS, SQS, etc.

And if you happen to be a high volume user, one chat to your account representative and you get very generous volume discounts across the board, saving you even more.

[sieabah]

What is almost never considered is that these large corporations don't pay public pricing. It's all negotiated to be heavily discounted. Bandwidth is hardly an issue like it is for regular customers.

[jlawer]

Not to mention Atlassian are moving bitbucket to be in the same cloud as their other services. They likely could have negotiated a cheaper per unit cost on their entire cloud workload, thus its possible this has generated a saving for Jira, confluence and all.

[idnefju]

It’s not for performance, atlassian performance degraded for us with this move.

I’m sure it was a good decision to save costs on their part though.

[fomine3]

For some service, they need to be on AWS because customers don't want to pay egress fee from AWS.