Should have switched to bare metal, things like AWS cost a ton more than you realize, it just looks like a good deal because you have smaller cost up front. But in the long run managing your own systems at places like Softlayer can save you a bundle.
EDIT: I don't care about downvotes, but if you are going to do it at least leave a comment of why you think I am wrong.
1. AWS saves us money despite the apparent higher costs because our developers spend more time working on our core business than managing infrastructure, and developer time is our most expensive resource.
2. Softlayer completely f'ed us twice now with down time resulting in a substantial dollar loss.
> 1. AWS saves us money despite the apparent higher costs because our developers spend more time working on our core business than managing infrastructure, and developer time is our most expensive resource.
But in most cases you lose any preserved savings because of the very ease of use and lack of needing to think about the infrastructure management. This is because more often than not problems are solved by just adding another instance, vs actually solving the larger problem.
Also, I would say part of the core business is providing the server, not simply writing the software that is the interface to the service being provided to the end users. E.G. Keeping servers up and and the right number of them in operating is just as important to the core of the business as writing the software / website of the service you are intending on providing. Outsourcing that a 3rd party removes your ability for you to control the entire service that you intend on providing, it is left in the hands of somebody else who may not be aligned to your business needs.
> 2. Softlayer completely f'ed us twice now with down time resulting in a substantial dollar loss.
AWS has f'ed near half the internet before, and probably will do so again.
Service outages are shitty, I get it, but my suggestion for Softlayer was just that, there are others, and my favorite is actually colo with everybody else's cheap EOL servers.
As a anatotal I have seen far to many projects move from bare metal, to AWS or the likes who at a minimum tripple the running cost for less performance. You could say the engineers suck, but I thought they were going to get to spend more time working on core business?
I have ran the numbers many time myself and I can't support AWS for anything other than bootstrapping a well funded startup. After that if you want to lower your cost you are either going to have to jump through hoops and formulate your service to use the cheapest parts all while not focusing on the busness but trying to get that AWS bill down.
Does part of your core business also involve mining silicon, crystalizing it, cutting the crystals into wafers, and then etching those wafers to make the chips your servers are built from?
Somewhere, you're drawing the line between what you build in-house and what you outsource. Where to draw the line depends on many factors. There's definitely no "right" place, such as never using cloud resources.
Vertical integration is function of size. At scale even the small efficiency improvements can make big differences, Facebook run their own DCs and design some hardware too coz it probably makes sense at their scale.
Vertical integration is a function of size and other variables.
For example, I used to work for a solar power company that would install panels on the roofs of big box stores and then sell the power to the stores below grid prices. This company did not manufacture panels -- they bought them at market rate from various suppliers.
The company got acquired by a major silicon wafer manufacturer. One of the things this wafer company made were solar panels -- boom, vertical integration. What a win, right? Well, as it turned out, after about a year, their competitors were able to bring their panel prices down whereas the acquiring company moved a bit more slowly. Thus the wafer company became an albatross around the solar power company's neck. They had to use the parent company's panels, instead of their more cost-effective competitors, and ended up no longer being competitive in the market.
So vertical integration can and often does make sense, but it comes with a loss of agility, and sometimes being nimble is more important.
While there are no fairy tale solutions . Vertical integration gives lot more control which can give gain competitive advantage if done correctly either by doing it cheaper or adding features your competiton cannot
In your ancedote the buying company seems to be benefit with increased sales for a non competitive product. Ofcourse they are doing at the expense of the eroding target , but they are deriving value not the way you see it perhaps or the best possible one but value nonetheless.
Developers managing infrastructure is expensive. Get some experienced architects and operations folks who dont let developers use resources because they are there.
Amazes me when I see automation stacks with 5x the hardware you need to run the entire production / dev / stage environments
> Developers managing infrastructure is expensive.
This is a phrase tossed around a lot, and I am challenging it.
If you make the developers part of the process then the software that is produced is forced to keep the entire vertical inline with the business needs.
You do this by giving the developers the budget along with the requirements. Developers are smart, they will figure out how produce the solution with the budget at hand. And when they realize they will be the one getting the call when things go wrong they will ensure systems are robust and fault tolerant by baking those features directly into the software they are building.
But I guess part of the process is there is a big effort to dumb down each segment of building software so people who have no desire to build software can simply use people as pawns in the game of business. It really shocks me that we still have businesses built with the core of people coming form antiquated business schools. Where they have reinforce the notion that a business needs to be swamped with a bunch of people who collect big paychecks for attending meeting after meeting after meeting while the engineers and developers continuously bring real value to organization.
Nobody ever talks about that CEO, CFO, or VP of whatever or Director of what ever as being expensive. Nobody ever talks about how to cut cost at that level. Meanwhile we have all sorts of strategies and cost cutting measures to lower the amount of budget going to the very people who are actually creating the value in a company.
I will tell you what is expansive. Paying some guy 6+ figures to tell you the guys who build the software that the company's revenue comes from is expansive. When was the last time you heard somebody outsourcing the executive team at a company ?
For those enraptures and founders out there, please don't get me wrong. I am not talking about you. I am talking about the people you may hire, or the people that other people say you need to hire.
> managing your own systems at places like Softlayer
The problem is, at a place like SL, they're not actually your own systems.
> can save you a bundle
It can, and usually does provide at least some savings, but it still comes with both the obvious form of vendor lock-in, and a subtler disadvantage, which is that you're limited by SL's specific hardware and engineering (including network) choices.
The real (performance and/or cost) magic of running ones own servers comes from custom combination of commodity ingredients, not merely lack of virtualization.
> The problem is, at a place like SL, they're not actually your own systems.
You are right, I do prefer colo and EOL hardware for low budget operations. But SL is a step lower than AWS. SL still can give you a punch in the face. But the point I really want is bare metal more than anything.
Also, if you need current gen stuff not owning can be nice because you won't have to repair when they break.
> It can, and usually does provide at least some savings, but it still comes with both the obvious form of vendor lock-in, and a subtler disadvantage, which is that you're limited by SL's specific hardware and engineering (including network) choices.
I advise you never to use vender provided features (that means avoiding AWS features along with depending on SL network topology (which is vlans, nothing special)).
Anybody making a service today should be able to buy a server at any provider, or setup a colo anywhere and join those resources in and get advantage of those resources instantly. All without depending on stacks and software you have no control over.
I hear this as a reason not to own, but I struggle to understand it, given my experience with how reliable modern hardware is (with notable exceptions[1]).
Regardless, you don't have to repair anything broken at all. It may seem preposterous at first, but, for low enough failure rates, on things with low enough replacement cost [2], using warm spares and abandoning in place can be a reasonable strategy. This can work remarkably well for disks in remote locations, especially in non-hot-swap enclosures.
> depending on SL network topology
One doesn't have a choice, though. There's no alternate topology (nor maximum bandwidths) available. Not taking maximum advantage may be premature optimization (optimizing for stability, rather than performance).
> Anybody making a service today should be able to buy a server at any provider, or setup a colo anywhere and join those resources in and get advantage of those resources instantly.
That would be nice, but it's tantamount to asking everyone go "multi-cloud", which removes too much of the "don't have to think/worry about it" purported benefit of cloud/IaaS.
[1] high-density (say, more than 1socket/U) seems to be the worst offenders
[2] by which I mean the current equivalent, including something like half of a server of a quarter of a CPU
> EDIT: I don't care about downvotes, but if you are going to do it at least leave a comment of why you think I am wrong.
That would defeat the entire purpose of using voting to manage the signal to noise ratio; in general, if a comment is worthy responding to, it shouldn't be downvoted, and vice versa.
I thought about this. Here is what I came up with.
If you upvote, and had to reply saying the same thing because you agree no new information is presented. Thus simply upvoting is needed.
But if you downvote, that means somebody has a differing opinion. Leaving it at a downvote only lets us know somebody disagrees and no new dialog can be had about what other ways something could be viewed.
That being said I think we mostly use the upvote and downvote feature on HN wrong but I could be wrong because I never really looked it up.
I think the intent is a upvote and downvote marks relevance, not agreement, but we often use it for agreement when upvoting. And downvotes seem to happen for both relevance and disagreements.
For example, replying to a post about what hosting provider is best with "I like chocolate ice cream", and "Rackspace is a good option" could both be downvoted for very difference reasons. One for being off topic, OR not liking chocolate ice cream" and the other for simply hating Rack Space.
So if you find you are downvoting something that has no comments that cover your settlement and the post is on topic then it would be helpful for many who may not share your sentiment if you explain why. I don't think that is too much to ask for.
You're downvoted for saying "You, person I don't know, should have used a different solution because of reason X" without considering they thought of that when they made their decision.
It was in the phrasing, not in some defense of the "Cloud", which is overrated for existing businesses.
If Azure costs 10x what bare metal costs, it's choosing between 1% of my tech budget or 0.1%. I doubt the multiplier is that bad, but bare metal isn't a consideration either way.
There's also some good integration with PM and IDE software that we use and enjoy.
EDIT: I don't care about downvotes, but if you are going to do it at least leave a comment of why you think I am wrong.
I look forward to the national election where a politician stops what they're doing to complain every single time someone doesn't vote for them, and interrupt what's happening to demand an explanation.
That isn't how voting works.
"cost a ton more than you realize" - unless you look at the prices for the contract you're agreeing to? Y'know, like you do?
"Should have" - telling someone what their company should have done, without even knowing what their company /is/, let alone what they were trying to achieve and what constraints they had on their available options. Assuming or implying that the cheaper thing is always the more correct thing and no other factors can change that.
I don't care about the upvotes either. I care about why somebody thinks one way or the other.
Maybe I will agree, or disagree. Maybe I will learn something. But simply downvoting because you don't agree leaves no new information for me or anybody else reading to learn something.
I could be horribly wrong. And the process of being told why somebody thinks I am wrong might inlighten somebody else even if I hold to my initial view.
That being said, the voting system AFAIK on HN is more about being relevant to the topic, and less about agreeing. Although in practice upvotes tend to be more about agreeing and down votes more about being off topic or illrevent.
Not everybody can downvote, so when somebody does it means they have been around long enough to gain enough magical internet points to do so. So they might have some real insight that could help a OP or inlighten others. Along with making a good conversation even if no agreement is made at the end.
> I look forward to the national election where a politician stops what they're doing to complain every single time someone doesn't vote for them, and interrupt what's happening to demand an explanation.
Nobody is a politician here, what does that have to do with anything? This entire site was designed to talk about things, so I am hardly interrupting "what's happening"
> "cost a ton more than you realize" - unless you look at the prices for the contract you're agreeing to? Y'know, like you do?
AWS has public pricing, I can compare them to other publicly priced places. I don't need a contract that simply don't exist in most cases.
> "Should have" - telling someone what their company should have done, without even knowing what their company /is/, let alone what they were trying to achieve and what constraints they had on their available options. Assuming or implying that the cheaper thing is always the more correct thing and no other factors can change that.
I can compare server for server , compute for compute, and all the above. I don't really need to know what our anybody else is doing with a system to understand the raw resources you can get from the services offered and compare them based off that.
Not everybody can downvote, so when somebody does it means they have been around long enough
And long enough to know that they can write a reply if they want to write a reply, without being prompted.
Nobody is a politician here, what does that have to do with anything? This entire site was designed to talk about things, so I am hardly interrupting "what's happening"
It's illustrative because politicians are the main thing people deal with when 'voting' for something, and the scale of political votes draws attention to the problem. Imagine HN if every comment that got downvoted had a line asking for an explanation - that would be almost every single comment - totally interrupting the flow of the comment threads. Imagine if political votes with millions of voters were stopped by someone demanding an explanation for one or two "down"votes. It would be unworkable. Voting is an aggregate, noisy, large-scale summary. And now this really is dragging it off-topic talking about voting instead, and entreaties to "explain your votes" to people who could have done that if they wanted to, always will tend towards "that's not downvote-worthy", "yes it is", "no it isn't".
AWS has public pricing, I can compare them to other publicly priced places.
Right. So why are you telling other people they are spending "more than they realize"? Either they didn't bother reading the prices, or they don't understand the prices - where's any evidence for either of those? How is it any more than an insult about their inability to understand the public pricing before signing up?
"I can compare server for server , compute for compute, and all the above. I don't really need to know what our anybody else is doing with a system to understand the raw resources you can get from the services offered and compare them based off that."
Indeed you can. What you can't do - or what I disagree with - is you saying "I can get cheaper compute resources here, /so/ that's what your company /should/ have done". It doesn't matter what you started with, who you employ, what skills they have, what constraints you had on timescale or rewriting or redesigning, what your management or investors would support, what discounts you could get, what your future plans are, none of those things matter, only a rudimentary price-per-ghz is enough to tell you that you did the wrong thing and say I know better.
EDIT: I don't care about downvotes, but if you are going to do it at least leave a comment of why you think I am wrong.