6 ms·

One can nit pick some of his points, but the general essay is right on point. Developers need to understand what they are building on top of to be able to navig

One can nit pick some of his points, but the general essay is right on point. Developers need to understand what they are building on top of to be able to navigate non-trivial problems. I found this article from Saturday worth the read, as it explains what Docker actually is, and why no one needs it: http://catern.com/posts/docker.html http://catern.com/posts/docker.html. The same goes for everyone working in the cloud; I've build my own clusters to compare the effort and expense (rather than just accept the marketing line) and running one's own servers, from one to hundreds, is exponentially more cost effective than any cloud service. People are not doing their homework, and the herd mentality owns the entire tech industry.

galois198 8y ago
Please could you elaborate on the cloud vs on-premise comparison? Are you including the initial on-prem costs of buying the hardware?
- bsenftner 8y ago
  Yep. The expense of the cloud is crazy; in most corporate situations the expense of purchasing the same servers used by the cloud service is 1/4 your monthly cloud bill. The markup is unethically high, so they've brainwashed an entire generation of young developers and their equality naive CTOs that this is anything other than a swindle. Running a server is child's play, people. Don't believe the hype!
  - mmt 8y ago
    People may have a strong reaction to the word "crazy", but I'm not sure what else to call it, other than "mass delusion" or "mass hallucination". I've done this analysis for (mostly small-scale) companies for about 10 years, and, every time, the cost of using public cloud computing has been 2x-10x of running ones own servers for 3 years. I've only had one exception, where the incremental cost of compute was 1:1, if the servers were placed in more expensive space near the other servers in the existing datacenter. Data transfer costs made cloud more expensive, of course. > Running a server is child's play, people. I'd say that's a bit of a hyperbole, but I agree that the incremental expertise (and time/effort) required to deal with hardware over cloud, is far from the prohibitve headache it's made out to be.
    - poulsbohemian 8y ago
      Much like in The Princess Bride, when companies say they are moving to the cloud for "cost," that word does not mean what they think it means. What they are really saying is that it is "cheaper" to swipe their corporate card for some AWS each month than it is for them to deal with the political and cognitive effort of doing it themselves. You know you can cook a better and cheaper meal at home but then you have a lot more effort than if you just go out to a restaurant and swipe your card. Same deal here, and most companies just want to get on with being a health care provider, energy company, financial services company, etc and they are willing to admit that they just don't have the stomach for IT. Going to the cloud is one of those places where they say cost but what they really mean is <networking | servers | power | etc> are _hard_.
      - mmt 8y ago
        > Going to the cloud is one of those places where they say cost but what they really mean is <networking | servers | power | etc> are _hard_. But I (and the GP) addressed that. It just isn't hard. Your earlier comment is closer to the truth: > political and cognitive effort of doing it themselves With "traditional" industries, it seems the problem is entirely political, and it actually is about cost, because cloud isn't competing against the commodity gear we've been talking about here. Cloud is competing against "enterprise" hardware, which easily costs even more than cloud, not least of which due to maintenance contracts. There was some discussion of this (somewhat tangential to the main topic) I participated in here https://news.ycombinator.com/item?id=17329028 https://news.ycombinator.com/item?id=17329028
  - davidjnelson 8y ago
    Well, unless your db master fails. Do yours automatically fail over?
    - bsenftner 8y ago
      Of course. That's essential to any enterprise.
jasonlotito 8y ago
> The same goes for everyone working in the cloud; I've build my own clusters to compare the effort and expense (rather than just accept the marketing line) and running one's own servers, from one to hundreds, is exponentially more cost effective than any cloud service. I have servers in the data center and in the cloud, and I can say that the line "is exponentially more cost effective than any cloud service" has so many caveats attached, it's not even worth debating. Simply put, if all you are doing is comparing running X servers in the DC and running the same servers in the cloud and that's it, yes, the cloud will be more expensive. This is why no cloud provider recommends simply doing that. > People are not doing their homework You should continue to do homework.
- karmakaze 8y ago
  > "I have servers in the data center and in the cloud, and I can say that the line" is exponentially more cost effective than any cloud service" has so many caveats attached, it's not even worth debating. Only thing I'll add is that at best it's asymptotic to a constant factor more cost effective.
cs02rm0 8y ago
I found this article from Saturday worth the read, as it explains what Docker actually is, and why no one needs it: http://catern.com/posts/docker.html.* http://catern.com/posts/docker.html.* Not sure I agree with the conclusions. A bunch of OS-specific alternatives don't compare as I see it, but regardless, for me the value of Docker isn't the filesystem trees alone but the tools that then leverage them; compose, swarm, etc.
edem 8y ago
Do you buy your own hardware? How do you run your own servers?
- bsenftner 8y ago
  Yes, I buy new servers for high compute purposes, and used ones for file shares and web heads. A full cabinet at a colo is around $800 a month, and that will hold 48 servers. 24 physical high compute servers in the cloud will run over $300K per month running 24/7. Owning your own? $800! And the savings that first month bought the fucking servers!
lalos 8y ago
The post you referenced reminded me of this HN comment regarding Dropbox right when it launched: https://news.ycombinator.com/item?id=9224 https://news.ycombinator.com/item?id=9224 It might be simple for an adept developer but not all companies can find/hire that kind of talent. You need abstractions that simplify dev ops in a way. Yes, docker does not solve everything and should only be used when you can leverage some of its pros.
twblalock 8y ago
> People are not doing their homework, and the herd mentality owns the entire tech industry. I've seen that, but I've also seen individual engineers and entire departments refuse to move on from what they are using. When they eventually realize they need to change, they are a lot further behind than they were initially, and it takes a lot more work to catch up. A lot of organizations are getting real value out of Docker (and Kubernetes and NoSQL and React and all of the other trendy new things), and saying that "no one needs it" is just as myopic and wrong as being part of the herd mentality.
- bsenftner 8y ago
  Docker adds unnecessary complexity, added insecurity, and the illusion that this is easier than just making a container by hand. It's insanity.
- mmt 8y ago
  > A lot of organizations are getting real value out of Docker (and Kubernetes and NoSQL and React and all of the other trendy new things) At the risk of nitpicking a word that may have been just used for emphasis, how can you be sure the value is real, though? Perceived value can be tough to tell apart, especially in a monoculture (called "herd mentality" earlier, perhaps too uncharitably). Also, even if the value is real, is it technical, or political/behavioral? If it's the latter, that doesn't minimize theri value, but let's not pretend their advantages are technical. We could, instead, discuss if using the new tool is the best (or only) way to effect that change. It may be. It may even be that which tool it is doesn't matter, so long as it's different, as has turned out to be the case in ergonomics and RSI. We've certainly seen some negative "NoSQL" headlines. How many startups used it to their eventual detriment, just following a trend and/or doing premature optimization? I'm not saying it can't provide value, rather than mere popularity and even perceived value, especially initially [1] > saying that "no one needs it" is just as myopic and wrong as being part of the herd mentality. I think that's what's called a "shallow dismissal" and is discouraged by the guidelines. The bare statement would be mere hyperbole and could possibly be "just as" wrong, but it was in reference to an entire article, which lays out the technical reasoning, which it's up to you to refute and point out they myopia of. [1] I am reminded of how "easy" MySQL replication was always touted to be.. except that people choosing it did so with an empty database. If it broke, with a far-from-empty database, I found myself wishing for one of PostgreSQL's much more "difficult" options.
ozim 8y ago
I can't see how I am going to get budget for "yeah guys I need half a year to provision 100 servers, 5 sys admins" VS "yeah I need 2 developers and a month, we can start selling to customer, just make markup adjusted with price for cloud storage". We have to minimize time to market, make money not build servers, but maybe others have other priorities.
- bsenftner 8y ago
  You're already operating from the cloud services perspective. Try 100 cloud servers is the same as 5 physical servers. Say 100 cloud servers runs your organization X dollar per month, plus the team of engineers to build out and operate there. For 1/4 of one month's cloud expense one can buy 5 servers that are the same hardware the cloud services are trickle renting you. With your month over month savings, you CAN afford those hard to find developers. Many of them are actually unemployed because everyone thinks they need to use the cloud - I've been able to pick them up for spot situations over my head for $15 an hour.
- mmt 8y ago
  > "yeah guys I need half a year to provision 100 servers, 5 sys admins" It's this kind of gross exaggeration that's keeping AWS so profitable, I suspect. Half a year and 5 sysadmins for only 100 server would do the NYC MTA subway construction team proud. > We have to minimize time to market, make money not build servers This is a false dichotomy, though. Of course you'd start with something where you can have infrastructure on day 1 of product launch. At small enough scale, cloud (or a VPS) is actually cheaper, too. Once you're at "time to market", if you're not re-evaluating, I'd say that's foolish. > make markup adjusted with price I'd expect any business to prefer to keep that markup for themselves as profit, assuming they even have that pricing power in the first place.

galois198 8y ago

Please could you elaborate on the cloud vs on-premise comparison? Are you including the initial on-prem costs of buying the hardware?

bsenftner 8y ago
Yep. The expense of the cloud is crazy; in most corporate situations the expense of purchasing the same servers used by the cloud service is 1/4 your monthly cloud bill. The markup is unethically high, so they've brainwashed an entire generation of young developers and their equality naive CTOs that this is anything other than a swindle. Running a server is child's play, people. Don't believe the hype!
- mmt 8y ago
  People may have a strong reaction to the word "crazy", but I'm not sure what else to call it, other than "mass delusion" or "mass hallucination". I've done this analysis for (mostly small-scale) companies for about 10 years, and, every time, the cost of using public cloud computing has been 2x-10x of running ones own servers for 3 years. I've only had one exception, where the incremental cost of compute was 1:1, if the servers were placed in more expensive space near the other servers in the existing datacenter. Data transfer costs made cloud more expensive, of course. > Running a server is child's play, people. I'd say that's a bit of a hyperbole, but I agree that the incremental expertise (and time/effort) required to deal with hardware over cloud, is far from the prohibitve headache it's made out to be.
  - poulsbohemian 8y ago
    Much like in The Princess Bride, when companies say they are moving to the cloud for "cost," that word does not mean what they think it means. What they are really saying is that it is "cheaper" to swipe their corporate card for some AWS each month than it is for them to deal with the political and cognitive effort of doing it themselves. You know you can cook a better and cheaper meal at home but then you have a lot more effort than if you just go out to a restaurant and swipe your card. Same deal here, and most companies just want to get on with being a health care provider, energy company, financial services company, etc and they are willing to admit that they just don't have the stomach for IT. Going to the cloud is one of those places where they say cost but what they really mean is <networking | servers | power | etc> are _hard_.
    - mmt 8y ago
      > Going to the cloud is one of those places where they say cost but what they really mean is <networking | servers | power | etc> are _hard_. But I (and the GP) addressed that. It just isn't hard. Your earlier comment is closer to the truth: > political and cognitive effort of doing it themselves With "traditional" industries, it seems the problem is entirely political, and it actually is about cost, because cloud isn't competing against the commodity gear we've been talking about here. Cloud is competing against "enterprise" hardware, which easily costs even more than cloud, not least of which due to maintenance contracts. There was some discussion of this (somewhat tangential to the main topic) I participated in here https://news.ycombinator.com/item?id=17329028 https://news.ycombinator.com/item?id=17329028
- davidjnelson 8y ago
  Well, unless your db master fails. Do yours automatically fail over?
  - bsenftner 8y ago
    Of course. That's essential to any enterprise.

jasonlotito 8y ago

> The same goes for everyone working in the cloud; I've build my own clusters to compare the effort and expense (rather than just accept the marketing line) and running one's own servers, from one to hundreds, is exponentially more cost effective than any cloud service. I have servers in the data center and in the cloud, and I can say that the line "is exponentially more cost effective than any cloud service" has so many caveats attached, it's not even worth debating. Simply put, if all you are doing is comparing running X servers in the DC and running the same servers in the cloud and that's it, yes, the cloud will be more expensive. This is why no cloud provider recommends simply doing that. > People are not doing their homework You should continue to do homework.

karmakaze 8y ago
> "I have servers in the data center and in the cloud, and I can say that the line" is exponentially more cost effective than any cloud service" has so many caveats attached, it's not even worth debating. Only thing I'll add is that at best it's asymptotic to a constant factor more cost effective.

cs02rm0 8y ago

I found this article from Saturday worth the read, as it explains what Docker actually is, and why no one needs it: http://catern.com/posts/docker.html.* http://catern.com/posts/docker.html.* Not sure I agree with the conclusions. A bunch of OS-specific alternatives don't compare as I see it, but regardless, for me the value of Docker isn't the filesystem trees alone but the tools that then leverage them; compose, swarm, etc.

edem 8y ago

Do you buy your own hardware? How do you run your own servers?

bsenftner 8y ago
Yes, I buy new servers for high compute purposes, and used ones for file shares and web heads. A full cabinet at a colo is around $800 a month, and that will hold 48 servers. 24 physical high compute servers in the cloud will run over $300K per month running 24/7. Owning your own? $800! And the savings that first month bought the fucking servers!

lalos 8y ago

The post you referenced reminded me of this HN comment regarding Dropbox right when it launched: https://news.ycombinator.com/item?id=9224 https://news.ycombinator.com/item?id=9224 It might be simple for an adept developer but not all companies can find/hire that kind of talent. You need abstractions that simplify dev ops in a way. Yes, docker does not solve everything and should only be used when you can leverage some of its pros.

twblalock 8y ago

> People are not doing their homework, and the herd mentality owns the entire tech industry. I've seen that, but I've also seen individual engineers and entire departments refuse to move on from what they are using. When they eventually realize they need to change, they are a lot further behind than they were initially, and it takes a lot more work to catch up. A lot of organizations are getting real value out of Docker (and Kubernetes and NoSQL and React and all of the other trendy new things), and saying that "no one needs it" is just as myopic and wrong as being part of the herd mentality.

bsenftner 8y ago
Docker adds unnecessary complexity, added insecurity, and the illusion that this is easier than just making a container by hand. It's insanity.
mmt 8y ago
> A lot of organizations are getting real value out of Docker (and Kubernetes and NoSQL and React and all of the other trendy new things) At the risk of nitpicking a word that may have been just used for emphasis, how can you be sure the value is real, though? Perceived value can be tough to tell apart, especially in a monoculture (called "herd mentality" earlier, perhaps too uncharitably). Also, even if the value is real, is it technical, or political/behavioral? If it's the latter, that doesn't minimize theri value, but let's not pretend their advantages are technical. We could, instead, discuss if using the new tool is the best (or only) way to effect that change. It may be. It may even be that which tool it is doesn't matter, so long as it's different, as has turned out to be the case in ergonomics and RSI. We've certainly seen some negative "NoSQL" headlines. How many startups used it to their eventual detriment, just following a trend and/or doing premature optimization? I'm not saying it can't provide value, rather than mere popularity and even perceived value, especially initially [1] > saying that "no one needs it" is just as myopic and wrong as being part of the herd mentality. I think that's what's called a "shallow dismissal" and is discouraged by the guidelines. The bare statement would be mere hyperbole and could possibly be "just as" wrong, but it was in reference to an entire article, which lays out the technical reasoning, which it's up to you to refute and point out they myopia of. [1] I am reminded of how "easy" MySQL replication was always touted to be.. except that people choosing it did so with an empty database. If it broke, with a far-from-empty database, I found myself wishing for one of PostgreSQL's much more "difficult" options.

ozim 8y ago

I can't see how I am going to get budget for "yeah guys I need half a year to provision 100 servers, 5 sys admins" VS "yeah I need 2 developers and a month, we can start selling to customer, just make markup adjusted with price for cloud storage". We have to minimize time to market, make money not build servers, but maybe others have other priorities.

bsenftner 8y ago
You're already operating from the cloud services perspective. Try 100 cloud servers is the same as 5 physical servers. Say 100 cloud servers runs your organization X dollar per month, plus the team of engineers to build out and operate there. For 1/4 of one month's cloud expense one can buy 5 servers that are the same hardware the cloud services are trickle renting you. With your month over month savings, you CAN afford those hard to find developers. Many of them are actually unemployed because everyone thinks they need to use the cloud - I've been able to pick them up for spot situations over my head for $15 an hour.
mmt 8y ago
> "yeah guys I need half a year to provision 100 servers, 5 sys admins" It's this kind of gross exaggeration that's keeping AWS so profitable, I suspect. Half a year and 5 sysadmins for only 100 server would do the NYC MTA subway construction team proud. > We have to minimize time to market, make money not build servers This is a false dichotomy, though. Of course you'd start with something where you can have infrastructure on day 1 of product launch. At small enough scale, cloud (or a VPS) is actually cheaper, too. Once you're at "time to market", if you're not re-evaluating, I'd say that's foolish. > make markup adjusted with price I'd expect any business to prefer to keep that markup for themselves as profit, assuming they even have that pricing power in the first place.