Live data from Hacker News

We handle 80TB and 5M page views a month for under $400

blog.polyhaven.com

311–320 of 426 posts

Re: We handle 80TB and 5M page views a month for under $400

#312

Earlier quoted context omitted.

You missed the point, it's "pay $20 real quick to Cloudflare if you happen to get popular". There is a generous free tier, and for years I have paid nothing to Cloudflare, but I serve 2GB total every month to visitors, YMMV.

No I didn’t miss the point, $20 is weeks worth of meal in majority of countries. People from those countries run successful blogs without paying a penny to cloudlfare. I serve much more than 2GB per month without all that on a dynamic site. Paying cloudlfare for that is like putting a tarp over a dumpster fire to quickly hide it. I would rather think why my static site fails like shit when it doesn’t even need server…

Interpreting ‘weeks’ as 2 weeks, the GDP necessary to call that weeks worth of meals is $520 dollars. There’s only like 6 countries with a GDP that low.

If you are only talking about food, it may stretch a bit further, but you’re still far away from the majority of countries.

Re: We handle 80TB and 5M page views a month for under $400

#313

Earlier quoted context omitted.

I invite you to do the calculations of serving 80tb of data serving from S3. Over 6k was my result but happy to be proved wrong.

What are you serving on a static site/ blog worth 80TB? I also said CDN makes sense for huge files. Please don’t nitpick.

4 kg of feathers is the same weight as 4 kg of lead.

Same thing applies to millions of 10kb files. Whether or not files are large is irrelevant to whether a CDN is a good idea.

If every file only ever gets requested once it may be pointless.

Re: We handle 80TB and 5M page views a month for under $400

#314
post #266

Earlier quoted context omitted.

As long as you treat servers as cattle, you can use Hetzner's own load balancer service and then you don't have a SPOF that you manage yourself. Their LBs are advertised as redundant / fault tolerant.

I don't know if it is possible to treat "Serverbörse" servers as cattle. They are all different. I know that k8s and Docker Swarm could in theory balance load between different machines, but never tried it in practice. But I had in practice some weird glitches with different CPU/motherboards/memory. Also, comment I was replying to mentioned a 48 euro budget, it is a price of a single server.

i think the whole point of vm/container is to enable "servers as cattle"

Re: We handle 80TB and 5M page views a month for under $400

#315
post #313

Earlier quoted context omitted.

What are you serving on a static site/ blog worth 80TB? I also said CDN makes sense for huge files. Please don’t nitpick.

4 kg of feathers is the same weight as 4 kg of lead. Same thing applies to millions of 10kb files. Whether or not files are large is irrelevant to whether a CDN is a good idea. If every file only ever gets requested once it may be pointless.

Makes sense, thats what I mean too. There is a use case for CDNs which do no make sense to pay for if its a static site. Which can be done much cheaper and for free most of the time. Unless you are running a google scale static site.

Re: We handle 80TB and 5M page views a month for under $400

#316
post #164
post #52

Earlier quoted context omitted.

I must be missing something, are you able to elaborate on how CentralizedPro Flare for $20 solves the OPs problems? I've accessed plenty of sites physically located in the USA from various EU countries, not to mention AUS and I didn't notice much difference as long as they had a good interconnect. Maybe not the best for YouTube, but there's only a few of those. Don't get me wrong, I see the appeal and fell for CF ear…

> Don't get me wrong, I see the appeal and fell for CF early on but not a fan anymore since they've grown into such a behemoth. > Absolutely power corrupts absolutely. This seems to capture the gist of the argument clearly. I hadn't realized cloudflare had grown large enough to fall into the "too big to not be evil" bucket already.

brace yourself before you read up on cf's origin

Re: We handle 80TB and 5M page views a month for under $400

#317

Earlier quoted context omitted.

Not including extra human cost in the analysis is just disingenuous. I think to manage private servers of that size you would need at least two extra experts totalling at least $20k/month.

> I think to manage private servers of that size you would need at least two extra experts totalling at least $20k/month. What? You set up the deployment once, and then you only need to touch it when things go horribly wrong, which is every couple of months, or to make minor quick tweaks and run some updates. Let's be generous, and say you need 10 h/month, which is about 1/16 of a person-month. And if things go horri…

> What? You set up the deployment once, and then you only need to touch it when things go horribly wrong, which is every couple of months

Right, which is exactly why people pay extra for cloud managed services.

If things are going “horribly wrong” every couple of months then you must necessarily be on call 24/7 and never take vacation or time away. In practice, you need at least two people to manage on-call coverage so you’re not completely uncovered if someone gets sick, decides to take vacation, wants to travel away from a computer and so on.

Re: We handle 80TB and 5M page views a month for under $400

#318

Earlier quoted context omitted.

Not including extra human cost in the analysis is just disingenuous. I think to manage private servers of that size you would need at least two extra experts totalling at least $20k/month.

What?! Maybe if you hire SV-skilled engineers on location in Sillicon Valley, but you can easily serve 2GB/second (on infra worth $2K/month) with one sys-admin dealing with it, and for way less than a whopping $10k/month.

> with one sys-admin dealing with it,

And if that one sys-admin wants to go on vacation? Or travel away from a computer? Or takes another job?

You can never have “just one” admin handling a server and being on-call 24/7.

Would you really want a job where you could never, ever be away from a computer because you’re the only on-call person? This doesn’t work.

Re: We handle 80TB and 5M page views a month for under $400

#319

Earlier quoted context omitted.

I don't know if it is possible to treat "Serverbörse" servers as cattle. They are all different. I know that k8s and Docker Swarm could in theory balance load between different machines, but never tried it in practice. But I had in practice some weird glitches with different CPU/motherboards/memory. Also, comment I was replying to mentioned a 48 euro budget, it is a price of a single server.

i think the whole point of vm/container is to enable "servers as cattle"

Yes, but "Serverbörse" machines are very non-uniform, which may be bad for load balancing. See yourself: https://www.hetzner.com/sb

Re: We handle 80TB and 5M page views a month for under $400

#320

Earlier quoted context omitted.

> Maybe not, but is the target audience that shills out $20/month really the type of people who have optimized their site to such an extent that shaving 50ms off the request latency by having your edge cache geolocated is really the type of thing that makes the difference? The common mistake is to pick a server geographically close to yourself, only access it from low-latency connections, and then assume that everyon…

> Consider the Photopea.com website. The author explained in a comment below that he spends $60/month to host the site without a CDN. Several of us loaded the site and it took 2.5 - 5.0 seconds to load This is a conclusion i am extremely doubtful of. Ping time new york tokoyo is about 180ms. So lets say as a worse case the ping time to the single server is 180ms (its probably not that bad), and lets say the latency t…

> In this scenario you would basically have to have 25 round trips to bring the site from 5 seconds to 1 second

You’re forgetting that the TCP protocol itself is bidirectional. High latency connections will have lower throughout, especially at the beginning of transmission, because the data isn’t literally just streaming in one direction.

Post reply on HN