3 ms·

I would also like to know about how they have optimized arc and it's underlying runtime from racket . They seem to have proven that arc can scale .

eliben 9y ago
Scale in what sense? I'm genuinely curious. Isn't this mostly a network/storage/memory bound problem?
- kuschku 9y ago
  Back in the day, HN was a single-core process on a single machine. Obviously, that’s an entirely different story than services distributed over clusters of hundreds of different machines.
  - qb45 9y ago
    190 days ago still was: https://news.ycombinator.com/item?id=12911086 https://news.ycombinator.com/item?id=12911086 Plus caching for logged-out visitors.
jacquesm 9y ago
It mostly proves that cloudflare can scale. Not that long ago Dan asked us to log out due to server load so that pages would be served from the cloudflare cache rather than that they had to be generated on the fly.
- j_s 9y ago
  Thanks to DanG and others for their work keeping HN running! (for the past 3+ years: https://news.ycombinator.com/item?id=7493856 https://news.ycombinator.com/item?id=7493856 ) -- HN Scalability https://news.ycombinator.com/item?id=13755673#13756819 https://news.ycombinator.com/item?id=13755673#13756819 (3 months ago, on Ask HN: Is S3 down?) > If you don't plan to post anything, would you mind logging out? Then we can serve you from cache. https://news.ycombinator.com/item?id=12909752#12911870 https://news.ycombinator.com/item?id=12909752#12911870 (6 months ago, on Donald Trump Is Elected President) > please log out unless you intend to comment, so we can serve you from cache > We buried the previous thread on this because, at almost 2000 comments, it was pegging the server. | https://news.ycombinator.com/item?id=12911042 https://news.ycombinator.com/item?id=12911042 https://news.ycombinator.com/item?id=9174869#9175457 https://news.ycombinator.com/item?id=9174869#9175457 (1 year ago, on Ask HN: Why the HN Nginx error for articles below 9M?) > We turned old articles off a few hours ago as an emergency measure because the site was being crawled aggressively and our poor single-core Racket process couldn't keep up. Previous solutions that have served us well until now, such as IP-address rate limiting, appear no longer to suffice. -- Source: https://hn.algolia.com/?query=author:dang%20cache&type=comment https://hn.algolia.com/?query=author:dang%20cache&type=comme...
- koolba 9y ago
  They still ask us to do that when the thread has lots of comments.
- kogir 9y ago
  As of a few years ago, HN could run fine without CloudFlare and just nginx. CF was mostly to try and obscure the server's IP, and to handle (rare) DDoS attacks.
- stephenr 9y ago
  As opposed to just, you know, using ESI so your page loads are always from the cache.

eliben 9y ago

Scale in what sense? I'm genuinely curious. Isn't this mostly a network/storage/memory bound problem?

kuschku 9y ago
Back in the day, HN was a single-core process on a single machine. Obviously, that’s an entirely different story than services distributed over clusters of hundreds of different machines.
- qb45 9y ago
  190 days ago still was: https://news.ycombinator.com/item?id=12911086 https://news.ycombinator.com/item?id=12911086 Plus caching for logged-out visitors.

jacquesm 9y ago

It mostly proves that cloudflare can scale. Not that long ago Dan asked us to log out due to server load so that pages would be served from the cloudflare cache rather than that they had to be generated on the fly.

j_s 9y ago
Thanks to DanG and others for their work keeping HN running! (for the past 3+ years: https://news.ycombinator.com/item?id=7493856 https://news.ycombinator.com/item?id=7493856 ) -- HN Scalability https://news.ycombinator.com/item?id=13755673#13756819 https://news.ycombinator.com/item?id=13755673#13756819 (3 months ago, on Ask HN: Is S3 down?) > If you don't plan to post anything, would you mind logging out? Then we can serve you from cache. https://news.ycombinator.com/item?id=12909752#12911870 https://news.ycombinator.com/item?id=12909752#12911870 (6 months ago, on Donald Trump Is Elected President) > please log out unless you intend to comment, so we can serve you from cache > We buried the previous thread on this because, at almost 2000 comments, it was pegging the server. | https://news.ycombinator.com/item?id=12911042 https://news.ycombinator.com/item?id=12911042 https://news.ycombinator.com/item?id=9174869#9175457 https://news.ycombinator.com/item?id=9174869#9175457 (1 year ago, on Ask HN: Why the HN Nginx error for articles below 9M?) > We turned old articles off a few hours ago as an emergency measure because the site was being crawled aggressively and our poor single-core Racket process couldn't keep up. Previous solutions that have served us well until now, such as IP-address rate limiting, appear no longer to suffice. -- Source: https://hn.algolia.com/?query=author:dang%20cache&type=comment https://hn.algolia.com/?query=author:dang%20cache&type=comme...
koolba 9y ago
They still ask us to do that when the thread has lots of comments.
kogir 9y ago
As of a few years ago, HN could run fine without CloudFlare and just nginx. CF was mostly to try and obscure the server's IP, and to handle (rare) DDoS attacks.
stephenr 9y ago
As opposed to just, you know, using ESI so your page loads are always from the cache.