Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Is a VPS overkill for a static site? - Page 2
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Is a VPS overkill for a static site?

2»

Comments

  • donlidonli Member

    @FHR said:
    Static site? Just throw it on BunnyCDN

    BunnyCDN costs a minimum of $10/year - don't you know where you are ?

  • FHRFHR Member, Host Rep

    @donli said:

    @FHR said:
    Static site? Just throw it on BunnyCDN

    BunnyCDN costs a minimum of $10/year - don't you know where you are ?

    I'm sorry, I totally forgot. The provider should pay customers instead.

    Thanked by 2sanvit poisson
  • defaultdefault Veteran
    edited June 2019

    @FHR said:

    @donli said:

    @FHR said:
    Static site? Just throw it on BunnyCDN

    BunnyCDN costs a minimum of $10/year - don't you know where you are ?

    I'm sorry, I totally forgot. The provider should pay customers instead.

    John F Kennedy said:

    Let every nation know, whether it wishes us well or ill, that we shall pay any price, bear any burden, meet any hardship, support any friend, oppose any foe to assure the survival and the success

  • uptimeuptime Member

    Ask not if a vps is overkill for a static site - ask if a static site is overkill for a vps!

  • donlidonli Member

    @FHR said:

    @donli said:

    @FHR said:
    Static site? Just throw it on BunnyCDN

    BunnyCDN costs a minimum of $10/year - don't you know where you are ?

    I'm sorry, I totally forgot. The provider should pay customers instead.

    Until then we have google's free VPS tier and free Cloudflare CDN.

  • If you're looking for something efficient and cheap Netlify or BunnyCDN (I agree that 10$/y minimum is a bummer but if you use their own storage it becomes more interesting!). If you want to have fun, take control over your crypto and shit like that grab a cheap VPS. As it's static you can even grab two super cheap VPSes (even NAT VPSes!) and setup a round robin that can act as a poor man HA setup (most browsers will try the second ip if the first fails).

    Super cheap shared hosting with CF free plan in front configured will full caching so that you serve everything from their edge nodes would work really, really well for a couple of bucks a year.

  • v3ngv3ng Member, Patron Provider
    edited June 2019

    @datanoise said:
    As it's static you can even grab two super cheap VPSes (even NAT VPSes!) and setup a round robin that can act as a poor man HA setup (most browsers will try the second ip if the first fails).

    Are you sure about that?

    Edit: gave it a try, that indeed seems to be possible! Didn't know that!

    Thanked by 1uptime
  • jsgjsg Member, Resident Benchmarker

    @sanvit said:
    Who uses IPv6 ? It's time for IPv8!

    That is wisdom!

    The ugly big problem of IPv6 is the fact that there are only enough IPs for each atom in a few universes. Nowadays some scientists are certain however that there is a very, very large number of parallel universes. And we certainly don't want to be egomaniacal racists who'll deny some atoms in some far away universes their fair share of IPs, right?

    So, let's be bold and go towards IPv32k right away. And it's cheaper too than hot chocolate for atoms feeling disenfranchised in some far away universe!

  • @jsg said:

    @sanvit said:
    Who uses IPv6 ? It's time for IPv8!

    That is wisdom!

    The ugly big problem of IPv6 is the fact that there are only enough IPs for each atom in a few universes. Nowadays some scientists are certain however that there is a very, very large number of parallel universes. And we certainly don't want to be egomaniacal racists who'll deny some atoms in some far away universes their fair share of IPs, right?

    So, let's be bold and go towards IPv32k right away. And it's cheaper too than hot chocolate for atoms feeling disenfranchised in some far away universe!

    There is no amount of memory/address space that human stupidity can't fill up.
    Could have added a few more bits for addressing. :)

    In my understanding: IPv6 subnet space is relatively large, compared to IPv4 and I wouldn't be surprised if, within our lifetime, it becomes "crowded".
    Not every single address could be used, but if every large ISP, or a government, takes large enough address space, plus private contractors...
    With today's equipment routing is more of a problem than address number, but within 40 years... wouldn't bet a lot on it.

    Electricity was used to replace the gas lights at the start. Now it's used for all sorts of stuff.
    In terms of computers, especially the Internet, we are now near the "light-bulb" phase - with no idea what uses the future will bring (not expecting anything bright, nor humane, but that's another topic).

    "No one will need more than 637 kB of memory for a personal computer"

  • NeoonNeoon Community Contributor, Veteran

    No, since Shared Hosting often especially the cheap ones, get into performance issues.

    Thanked by 1huckleberrypie
  • jsgjsg Member, Resident Benchmarker
    edited June 2019

    @bikegremlin said:
    In my understanding: IPv6 subnet space is relatively large, compared to IPv4 and I wouldn't be surprised if, within our lifetime, it becomes "crowded".
    Not every single address could be used, but if every large ISP, or a government, takes large enough address space, plus private contractors...
    With today's equipment routing is more of a problem than address number, but within 40 years... wouldn't bet a lot on it.

    FYI: IPv6 allows for 2 ^ 128 IPs. That equates to about 2 ^ 95 IP addresses for each and every person on this planet which again is about 2 ^ 63 complete IP4 internet address spaces for each and every person on this planet. Or, to make it easier for humans who don't think binary but 10-based -> About 1.000.000.000.000.000.000.000 complete IP4 address spaces for each and every human. I think we can agree that that is not something that will be urgently needed anytime soon, "soon" as in "the next 10.000 years".

    And no whether a new IP version has 128 or 64 bits is not unimportant because 64 bits is the width of modern processors that is, all IP calculations can be simply and efficiently performed by todays processors - but IPv6 addresses can not and are an immense waste of resources and energy and will require by far more expensive equipment.

    And what for? Because some braindead imbeciles "thought" that half of todays total IP address space for each and every person on this planet wouldn't be enough. Because, you know, maybe mankind suddently grows a couple of of ten thousand times which then would leave everyone with, oh, gosh, only 65.000 IP addresses ...

    And low and behold, IPv6 did not take off big time although that was propagandized and pushed for multiple times since about a decade. Because those who had to actually carry the load and bear the costs and do the investments, like for example ISPs and providers, did recognize that IPv6 dramatically increases costs and doesn't offer a solution that even remotely delivers what is actually and reasonably needed. And there I'm not even talking about tons and tons of software needing to be changed or even rewritten ...

    To quickly shoot down another trendy argument for IPv6 -> IoT. That's BS. The vast majority of IoT is closely linked to the home and could - and should - hence be NATed and/port forwarded into private adress space.

    In summary IPv6 is like a mid size city of 50.000 citizens planning to build a hospital for 200 billion people, just in case ... It's utter idiocy, simple as that.

    Thanked by 1bikegremlin
  • @Neoon said:
    No, since Shared Hosting often especially the cheap ones, get into performance issues.

    Been there, done that. It's either you're restrained from a certain version of PHP or file uploads are constrained.

  • defaultdefault Veteran
    edited June 2019

    @jsg said:
    In summary IPv6 is like a mid size city of 50.000 citizens planning to build a hospital for 200 billion people, just in case ... It's utter idiocy, simple as that.

    We need to think ahead, we need to be creative. This is just an addressing system, still extremely simplistic and easy to calculate with today's computing power. With all this in mind, we still need to think ahead in aspects like:

    • teleportation where billions of cells need to have addresses and be placed back in a specific order;

    • atoms management in physics where you have to give each one a private address when attempting to control, like you already mentioned by throwing the idea into brainstorming;

    • addressing in universe: location on planet, to star systems, to galaxies, to parallel universes if those exists (currently is just a theory).

    • whatever exists beyond what Hubble telescope can analyze, even beyond cosmic microwave background.

    We are still just scratching the surface of our solar system, we may be way too early in thinking about capturing all the energy from photons of our sun, but we will need address space just to allow ourselves to think about the future, and not have so many issues within our involvement.

    Thanked by 1uptime
  • donlidonli Member

    The big error in IPv6 was lack of backward compatibility...

    "The lack of real backwards compatibility for IPv4 was the single critical failure," says Leslie Daigle, Chief Internet Technology Officer for the Internet Society. "

    https://www.networkworld.com/article/2265836/biggest-mistake-for-ipv6--it-s-not-backwards-compatible--developers-admit.html

  • jsgjsg Member, Resident Benchmarker

    I get that you are joking but I'll pick out 1 point that is a wide spread misunderstanding and actually dramatic.

    @default said:

    @jsg said:

    ...still extremely simplistic and easy to calculate with today's computing power.

    Nope. IPv6 addresses are four times the size of IP4 addresses. That also means that IP4 can be processed by 32 bit processors - which is exactly what's happening in reality. To use a 64-bit processor adds cost and complexity to design and manufacturing as well as higher power consumption. All of those are relevant reasons for most IP processing equipment (e.g. routers) to use 32 bit cores (also in ASICs).

    128-bit on the other hand would either need new processor generations or using special large registers and special instructions (e.g. AVX) which obviously increases cost and power consumption significantly.

    As for "easy to calculate", well, depends on from where you look. From a normal users perspective everything that needs less than say 1 microsecond ist "damn fast". From a routers perspective however a 10 times increase in processing time is very, very significant (a 10 times increase is a realistic ballpark figure for what we talk about).

    But there is another point: All those billions and billions of addresses must be switched, forwarded, routed too. So with IPv6 we ask the core equipment to do much more work while at the same time we very significantly increase the work load per address.

    Also note that yester-years 40 Gb/s ("really big!") capacity today is 200 Gb/s (meaning that's what's increasingly considered as "big capacity" e.g. in switches) - while the MTU stays the same which translates to "many, many more packets" are coming and needing to be switched, forwarded, and routed.

    And again: All that trouble for nothing but a wet and insane dream! Replacing IP4 with a far more sensible 64-bit system would avoid virtually all the above mentioned problems, at least to a large degree, and still offer half of todays total IP4 address space to each and every human. Assuming that for 99+% of humans "only" 65k IP addresses would be more than they could possibly use, we would still be well prepared with 64-bit addresses for a 100+ fold increase of global population (or, in other words, for a population that would be unsustainable for diverse other reasons anyway).

    Thanked by 1bikegremlin
  • bikegremlinbikegremlin Member
    edited June 2019

    @jsg said:

    @bikegremlin said:
    In my understanding: IPv6 subnet space is relatively large, compared to IPv4 and I wouldn't be surprised if, within our lifetime, it becomes "crowded".
    Not every single address could be used, but if every large ISP, or a government, takes large enough address space, plus private contractors...
    With today's equipment routing is more of a problem than address number, but within 40 years... wouldn't bet a lot on it.

    FYI: IPv6 allows for 2 ^ 128 IPs. That equates to about 2 ^ 95 IP addresses for each and every person on this planet which again is about 2 ^ 63 complete IP4 internet address spaces for each and every person on this planet. Or, to make it easier for humans who don't think binary but 10-based -> About 1.000.000.000.000.000.000.000 complete IP4 address spaces for each and every human. I think we can agree that that is not something that will be urgently needed anytime soon, "soon" as in "the next 10.000 years".

    Not every appliance will take just one IP address. That's what I meant by "large subnet space". That is: a company could get a hefty chunk of IP address space and actually use only 1% for actual appliances. But it's still taken.

    And no whether a new IP version has 128 or 64 bits is not unimportant because 64 bits is the width of modern processors that is, all IP calculations can be simply and efficiently performed by todays processors - but IPv6 addresses can not and are an immense waste of resources and energy and will require by far more expensive equipment.

    I agree - it's what I meant with "routing is more of a problem than address number". With today's equipment.

    And what for? Because some braindead imbeciles "thought" that half of todays total IP address space for each and every person on this planet wouldn't be enough. Because, you know, maybe mankind suddently grows a couple of of ten thousand times which then would leave everyone with, oh, gosh, only 65.000 IP addresses ...

    I understand your arguments, they are quite logical and sensible.

    Some 5 years ago I did a research on IPv6 to IPv4 comparison and transition ("translation?"). And I remember my conclusion from that time (which wasn't the part of the research, IPv6 is what it is, can't change it) was that they should have added 2 more bits. Unfortunately, I wasn't smart enough to write that down as well ("wise man writes down, a fool remembers" is a local saying). If memory serves me - it boiled down to using subnet groups for addressing various geo locations, then making smaller subnet "sub-groups" for countries, regions, down to company levels.

    What I certainly hadn't considered at the time, were the extra computing resources needed to implement that - which was probably the reason for stopping at the current number of address bits. Also, I'm far from convinced that my conclusion was correct/justified one. But based on previous experience (with CPU power, RAM, storage space...), I wouldn't be amazed if we manage to make available IPv6 address space scarce.

    And low and behold, IPv6 did not take off big time although that was propagandized and pushed for multiple times since about a decade. Because those who had to actually carry the load and bear the costs and do the investments, like for example ISPs and providers, did recognize that IPv6 dramatically increases costs and doesn't offer a solution that even remotely delivers what is actually and reasonably needed. And there I'm not even talking about tons and tons of software needing to be changed or even rewritten ...

    Agreed.

    To quickly shoot down another trendy argument for IPv6 -> IoT. That's BS. The vast majority of IoT is closely linked to the home and could - and should - hence be NATed and/port forwarded into private adress space.

    In summary IPv6 is like a mid size city of 50.000 citizens planning to build a hospital for 200 billion people, just in case ... It's utter idiocy, simple as that.

    I would assume one point of IPv6 is to avoid the need for any NAT use, at some point in the future when IPv4 is abandoned. However, that is where the bottleneck lies IMO - without using NAT, each appliance would use at least one IP address. With countries and companies taking large "chunks" of the available address space - "just in case, there is plenty".

    Still, I agree with you - it will probably still be enough... but I wouldn't be amazed if it ends up being exhausted.

    For one, no one had considered address space for planet earth, solar system... :) Nanobots and augmentation implants of each human - will they be forced to use "the primitive NAT" in order to be remotely controlled, or is everyone's birth right to get a decent size of IP address space? :) What about pets and wildlife - will they be roaming free without any surveillance?! The last, but not the least: can you imagine the number of idling VPS-s of LET members in 20 years time?! :)

    Thanked by 1default
  • ClouviderClouvider Member, Patron Provider

    Routing is a problem with Ipv6? Today? You gotta be kidding me guys.

  • jsgjsg Member, Resident Benchmarker

    @Clouvider said:
    Routing is a problem with Ipv6? Today? You gotta be kidding me guys.

    No, not today. But it certainly would be with well filled 40, 100, 200 Gb/s pipes. That's simply a technical fact.

    @bikegremlin said:
    Not every appliance will take just one IP address. That's what I meant by "large subnet space". That is: a company could get a hefty chunk of IP address space and actually use only 1% for actual appliances. But it's still taken.

    That's a completely different problem. And although even 64bit IPs would have enough address space for large organisations to get 24-bit subnets (what today is a/8), all requests for more than say 128 IPs should be justified and verified as a principle (keep in mind how we got into problems in the first place: by lots of organisations hording unjustifiably large spaces).

    Technically speaking, however there's no problem. Everyone really needing large spaces (incl. solid reserves) could easily get his IPs. Again: a 64-bit address space means that there would be half of todays IP4 space for each and every human on this planet.

    If memory serves me - it boiled down to using subnet groups for addressing various geo locations, then making smaller subnet "sub-groups" for countries, regions, down to company levels.

    That could be done with 64 bit IPs as well.

    I would assume one point of IPv6 is to avoid the need for any NAT use, at some point in the future when IPv4 is abandoned.

    You are right. That was indeed one important consideration - and an utterly imbecile one. Explanation: Today we use NAT to cope with IP scarcity. But something like NAT must be understood correctly. The basic mechanism is to seperate public and private IP space. And that, private IP space, is something highly desirable even with 128 bit addresses. A mechanism like NAT makes us users the masters of our private little universe and we do not want all of our devices to be in publicly routable address space.

    Thanked by 1bikegremlin
  • ricardoricardo Member
    edited June 2019

    Clouvider said: Routing is a problem with Ipv6? Today? You gotta be kidding me guys.

    Nothing is a problem if you throw enough resources at it. Was it a rhetorical question?

    I think JSG threw enough on the processor size angle. Even as a provider and it being 'not a problem' there is the issue of power consumption. And address sizes being orders of magnitude bigger and being sent in the trillions, daily, it's worthy of consideration.

    Using IPv6 seems like a be all and end all solution, given how many addresses it covers. 2^64 seems like it'd have at least covered mankind for a few hundred years. 2^128 seems like it'd take care of the human race and any other intelligent race in the galaxy.

    That said, bugger all to do with the OP.

  • ClouviderClouvider Member, Patron Provider

    @ricardo said:

    Clouvider said: Routing is a problem with Ipv6? Today? You gotta be kidding me guys.

    Nothing is a problem if you throw enough resources at it. Was it a rhetorical question?

    I think JSG threw enough on the processor size angle. Even as a provider and it being 'not a problem' there is the issue of power consumption. And address sizes being orders of magnitude bigger and being sent in the trillions, daily, it's worthy of consideration.

    Using IPv6 seems like a be all and end all solution, given how many addresses it covers. 2^64 seems like it'd have at least covered mankind for a few hundred years. 2^128 seems like it'd take care of the human race and any other intelligent race in the galaxy.

    That said, bugger all to do with the OP.

    Given it took over a decade for IPv6 to take off I’d rather have too many IPs than face another family change.

    It’s a done deal. It’s salt or die. There’s no point to discuss alternatives now.

  • ricardoricardo Member
    edited June 2019

    Clouvider said: There’s no point to discuss alternatives now.

    Agreed, though the merits of using a 64-bit address still stand. IPv6 is convenience but convenience comes at a price - and wastage especially in today's world (IMO) should never be ignored.

  • jsgjsg Member, Resident Benchmarker
    edited June 2019

    @Clouvider said:
    It’s a done deal.

    Is it really? I see shockingly little uptake although proponents push it really hard since years. And I see more and more doubts and self-criticism from what were former proponents or even ones who still are. A "done deal"? Certainly not! At least not in reality where there are still even ISPs and providers of significance who do NOT offer IPv6.

    And no that's not a philosophical discussion. It's as business as it can get. Because it's about "what do I get and how much do I pay for it?", a very classical business question indeed.

    What you get with IPv6 is "ridiculously too much - plus hefty inconvenience" and the price (complexity, computation, energy, etc) is ridiculously high.

    In other words: it's an about as ridiculously poor business suggestion as business suggestions can get. It's as if Pizza Hut offered "a pizza with half a mile radius - and you must eat it with chopsticks".

    There is NO 128-bit architecture of any significance and NONE of the major manufacturers even plans to build 128-bit processors anytime soon. Similarly as far as is known even the likes of Google (who occasionally do strange things for "political reasons") are NOT working on 128-bit ASICS nor do Juniper or Cisco.

    Sure, one can do it with 32-bit processors (which still are standard in todays equipment) but even with 64-bit processors (which are still not standard on the network plane) the costs will increase dramatically and performance wil DEcrease dramatically.

    Funny side note: Those who "take it easy" and see no problems in their equipment are actually based on the LOW uptake of IPv6. Of course todays equipment can handle some minor IPv6 traffic. But try that with filled 40 Gb/s, let alone 200 Gb/s pipes ...

    Thanked by 1bikegremlin
  • @jsg said:

    You are right. That was indeed one important consideration - and an utterly imbecile one. Explanation: Today we use NAT to cope with IP scarcity. But something like NAT must be understood correctly. The basic mechanism is to seperate public and private IP space. And that, private IP space, is something highly desirable even with 128 bit addresses. A mechanism like NAT makes us users the masters of our private little universe and we do not want all of our devices to be in publicly routable address space.

    Agree with the rest of the post.

    As for this:
    Yes, you should be able to use NAT when you want to - but not because you have to.

    Another thing is whether you will be allowed to. Today, you can't do every payment using a VPS, for example, for all sorts of (automated) checks. I wouldn't be surprised if, in the near future, we are forced to use "our" IPs for any Internet use, so it is all checked, monitored and recorded. It all seems to be going in that direction, in spite of the GDPR ("who would mind something like that, you aren't doing anything illegal..."). Paper money is already not allowed for larger payments in many countries. But that's another topic.

  • @Clouvider said:
    Routing is a problem with Ipv6? Today? You gotta be kidding me guys.

    What you work with for embedded networking gear is not the same as small, embedded IoT type networking gear that ran IPv4 stacks on kilo/megabytes of flash and ram for decades.

Sign In or Register to comment.