John Kristoff

UIC PhD candidate | Dataplane.org | Netscout. Internet infrastructure (#BGP, #DNS) and #infosec. Bit mechanic. Also: #Blues / tfr / #fedi22

2026-01-30

@b0rk The benefit is essentially the same as why you might want 64GB or RAM instead of 32GB. Throwing resources at a problem.

That wasn't the initial reason, that's just what they got for being one of the first. The edu I was at got a /16 even though enrollment was about 15,000. But space gets used inefficiently, it's hard to change, growth happens, NAT sucks :-), cheapest and easiest way out? Get more. And we did.

That seems to be what Amazon is still doing. In a rough measure I took recently, they've transferred in about 175 million IP4 addresses over the years. Thats where a lot of the big address movement action is these days.

2026-01-30

Our Weekend Reads and The Internet Last Week summary posts will be on vacation this weekend.

2026-01-30

@openfactory @paul_ipv6 @jschauma Another mistake I made was a homework assignment to use PGP to send me an encrypted message. I know, I know it's PGP, but there were numerous problems unrelated to PGP I never considered when asking students to install software. That was a bucket of cold water to the face.

2026-01-30

RE: flipboard.com/@newyorktimes/wo

Announced on Twitter. Would have been cooler if you started with that.

2026-01-29

@jschauma One of the most humbling experiences of my career was when I taught my first grad class. One particularly memorable moment was the first midterm exam I created. I showed it to a colleague who had taught classes before and his reaction was "Yeeha!! Now that is an exam!"

Every single student took the full 3 hours allowed on it. I was getting a few looks and hearing a few laborious releases of breath. A couple just gave up and left either very unhappy with me or (regretably) themselves.

It's really hard to understand how hard it is to design and deliver good lectures, homework, labs, and tests unless you've done it.

2026-01-29

@efscher I'm not sure how old this is, probably from the early 1990s. Not sure it'll protect anything from a surge anymore, but keeps things powered OK.

2026-01-29

Am I the only one still using these things?

#ThrowbackThursday

8-port (US) isobar power bar
2026-01-29

A Spanish hosting provider that had stuck with Path.net (#AS396998) for add-on DDoS mitigation services for select IP addresses/prefixes longer than most has had enough (translated from Spanish):

"Unfortunately, the Path.net service was discontinued due to recurring outages."

John Kristoff boosted:
2026-01-28

🚨New blog post🚨

My latest for @kentikinc updates last year's analysis on the exodus of IPv4 from Ukraine and covers the BGP routing cleanup at AT&T that removed numerous foreign routes previously originated by AS7018.

kentik.com/blog/from-ukraine-t

2026-01-28

Fasthosts (IONOS hosting and reseller brand in the UK):

"4.95% price increase across all products"

takes effect March 1, 2026.

2026-01-27

On May 1, 2026, #Google CDN Interconnect, Direct Peering, and Carrier Peering prices go up by as much as 2x in one region (i.e., NA $0.04 -> $0.08 / GiB).

cloud.google.com/consulting/cl

John Kristoff boosted:
Jacobo Nájerajacobonajera
2026-01-27

The brief restoration of internet access in Iran is causing spikes in the use of the Tor Network’s Snowflake pluggable transport. These changes are being reflected in Tor network metrics.

Graph of Bridge users from Iran
2026-01-26

@hardaker I like the testimonial :-)

2026-01-26

It's not 7018, but a noteworthy addition to the growing community of ASPA users:

social.bgp.tools/@newaspa/stat

#RPKI

2026-01-24

#infosec and #networking friends local and abroad, I hope this captures in part what some of you are fearing, sensing, or seeing. There is a lot more one could say, but topically relevant here, an excerpt to locally elected government officials:

"[...] I'm seeing and experiencing a level of disdain, backlash, and general distrust international colleagues have for the U.S I've never seen so widely held before. This is starting to creep into the ability for me and the community to work together. Trust is being broken, relationships are fraying, and technology itself is diverging at regional boundaries."

John Kristoff boosted:
Erik Nygren :verified:nygren@hachyderm.io
2026-01-23

@markd @becomingwisest @jschauma #Akamai has made #IPv6 dualstack the default for around a decade for new customer configurations. I talk some about why customers don't dual-stack (and why they should!) in akamai.com/blog/trends/10-year

Many customer sites have been on long enough to pre-date the switch of the default, but we do still see some rate of opting out. Some top reasons include:

* Operationally conservative and lack of understanding of the value ("if it isn't broke don't fix it" or "seems safer"), or have trouble justifying the effort to make the change safely with testing.
* Lack of support in their log processing and/or header processing and/or advanced business logic for IP handling. For example, companies may be using a third-party anti-fraud product which doesn't support IPv6. With Akamai's large base of Enterprise customers this is frustratingly common and the "just hash the IPv6 to an IPv4" makes things worse rather than helping.
* Auth tokens bound to IP addresses. (eg, setting it via IPv4 and validating via IPv6 fails, or vice-versa -- see akamai.com/blog/developers/why )
* Custom or uncommon clients which break in the presence of IPv6. This is rare but I've seen it, for example some local streaming provider may have a user base with old Smart TVs that interact poorly with some ISP's somewhat broken IPv6 setup. I hear this less now, but it used to be common even a few years ago.

Note that it's worth checking both the "www" and apex name version of some of these, as in some cases the apex example.com name is IPv4-only but redirects to a dualstack "www" name due to CNAME-at-the-apex limitations.

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst