Just like there's models for servers "small count of big iron" vs "large count of ephemeral VMs" there exists two models for Internet infra.
-
-
Replying to @patio11
As originally conceived, the Internet was a cloud of horrifically underresourced side projects stuck together. Very resilient to losing one.
2 replies 18 retweets 32 likes -
Replying to @patio11
As it presently exists, there are (in many verticals) a few players which are greatly concentrated. Losing any one hurts the network a lot.
1 reply 4 retweets 11 likes -
Replying to @patio11
If AppAmaGooBookSoft get a cold, much of the Internet (as perceived by end users -- humans or companies) ceases to function.
1 reply 2 retweets 8 likes -
Replying to @patio11
Geeks occasionally say "This would be better if we were still a widely distributed system!" but that is not obviously true.
3 replies 0 retweets 7 likes -
Replying to @patio11
An argument exists that concentration risk is acceptable because the costs of building reliable systems are *astronomical*.
2 replies 3 retweets 7 likes -
Replying to @patio11
SMTP lets you run an email server for $1 but achieving delivery for it independently costs $1+ million a year because spam is *that bad.*
1 reply 0 retweets 16 likes -
Replying to @patio11
You can write a web browser. You can write a web browser securely. It costs $100 million a year.
2 replies 5 retweets 16 likes
These thoughts were occasioned by the DNS outage today, but they come up every time we get browntime at a systemically important provider.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.