Hacker News new | past | comments | ask | show | jobs | submit login

Stanford appears to still use a technique I like to use on the LANs I control, too.

Putting different sites on different ports, using port numbers instead of hostnames.

Even on the internet, ports over 49151 are not protocol-specific; they are not assigned by IANA^1

For example, on port 8090

and on port 8091

http is noticeably faster than https.

Stanford is still providing http in addition to https.

Here, the http url has been submitted to HN, not the https one.

1. https://www.iana.org/assignments/service-names-port-numbers/...

Years ago, at a community college I worked for, we used a software stack that used misc. non-standard port numbers.

Rural households using the satellite Internet providers HughesNet and Wild Blue were unable to access the affected sites. In addition, the firewalls at the state prison and the local National Guard base also blocked access.

One of the issues for our affected students was that they could not register nor add/drop classes.

Just an anecdote to suggest that there may be undesirable consequences to using non-standard ports for a website.

What are the "undesired consequences" for Stanford. These sites are both alternatively available over HTTPS using port 443

Consider websites on a LAN under you control vs websites on the internet that is not under your control

Using a L2 overlay one create LANs over the internet. Small private networks. Personal use, not public. Recreational use, not commercial. Then one can host daemons, including httpds and websites, on the LAN using whatever ports she wants.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact