jm + networks   10

'Jupiter rising: A decade of Clos topologies and centralized control in Google’s datacenter networks'
Love the 'decade of' dig at FB and Amazon -- 'we were doing it first' ;)

Great details on how Google have built out and improved their DC networking. Includes a hint that they now use DCTCP (datacenter-optimized TCP congestion control) on their internal hosts....
datacenter  google  presentation  networks  networking  via:irldexter  ops  sre  clos-networks  fabrics  switching  history  datacenters 
october 2016 by jm
NA Server Roadmap Update: PoPs, Peering, and the North Bridge
League of Legends has set up private network links to a variety of major US ISPs to avoid internet weather (via Nelson)
via:nelson  peering  games  networks  internet  ops  networking 
january 2015 by jm
Juniper Adds Puppet support
This is super-cool.

'Network engineering no longer should be mundane tasks like conf, set interfaces fe-0/0/0 unit o family inet address 10.1.1.1/24. How does mindless CLI work translate to efficiently spent time ? What if you need to change 300 devices? What if you are writing it by hand? An error-prone waste of time. Juniper today announced Puppet support for their 12.2R3,5 JUNOS code. This is compatible with EX4200, EX4550, and QFX3500 switches. These are top end switches, but this start is directly aimed at their DC and enterprise devices. Initially, the manifest interactions offered are interface, layer 2 interface, vlan, port aggregation groups, and device names.'

Based on what I saw in the Network Automation team in Amazon, this is an amazing leap forward; it'd instantly render obsolete a bunch of horrific SSH-CLI automation cruft.
ssh  cli  automation  networking  networks  puppet  ops  juniper  cisco 
august 2013 by jm
Communication costs in real-world networks
Peter Bailis has generated some good real-world data about network performance and latency, measured using EC2 instances, between ec2 regions, between zones, and between hosts in a single AZ. good data (particularly as I was looking for this data in a public source not too long ago).

I wasn’t aware of any datasets describing network behavior both within and across datacenters, so we launched m1.small Amazon EC2 instances in each of the eight geo-distributed “Regions,” across the three us-east “Availability Zones” (three co-located datacenters in Virginia), and within one datacenter (us-east-b). We measured RTTs between hosts for a week at a granularity of one ping per second.


Some of the high-percentile measurements are undoubtedly impact of host and VM behaviour, but that is still good data for a typical service built in EC2.
networks  performance  measurements  benchmarks  ops  ec2  networking  internet  az  latency 
may 2013 by jm
paperplanes. Monitoring for Humans
A good contemplation of the state of ops monitoring, post-#monitorama. At one point, he contemplates the concept of automated anomaly detection:
This leads to another interesting question: if I need to create activity to measure it, and if my monitoring system requires me to generate this activity to be able to put a graph and an alert on it, isn't my monitoring system wrong? Are all the monitoring systems wrong? [...]

We spend an eternity looking at graphs, right after an alert was triggered because a certain threshold was crossed. Does that alert even mean anything, is it important right now? It's where a human operator still has to decide if it's worth the trouble or if they should just ignore the alert. As much as I enjoy staring at graphs, I'd much rather do something more important than that.

I'd love for my monitoring system to be able to tell me that something out of the ordinary is currently happening. It has all the information at hand to make that decision at least with a reasonable probability.


I like the concept of Holt-Winters-style forecasting and confidence bands etc., but my experience is that the reality is that anomalies often aren't sufficiently bad news -- ie. when an anomalous event occurs, it may not indicate an outage. Anomaly detection is hard to turn into a reliable alarm. Having said that, I have seen it done (and indeed our team has done it!) where there is sufficiently massive volume to smooth out the "normal" anomalies, and leave real signs of impact.

Still, this is something that Baron Schwartz (ex-Percona) has been talking about too, so there are some pretty smart people thinking about it and it has a bright future.
monitoring  networks  holt-winters  forecasting  confidence-bands  anomaly-detection  ops  monitorama  baron-schwartz  false-positives 
march 2013 by jm
drug cartel-controlled mobile comms networks
“The Mexican military has recently broken up several secret telecommunications networks that were built and controlled by drug cartels so they could coordinate drug shipments, monitor their rivals and orchestrate attacks on the security forces. A network that was dismantled just last week provided cartel members with cellphone and radio communications across four northeastern states. The network had coverage along almost 500 miles of the Texas border and extended nearly another 500 miles into Mexico’s interior. Soldiers seized 167 antennas, more than 150 repeaters and thousands of cellphones and radios that operated on the system. Some of the remote antennas and relay stations were powered with solar panels.”
mexico  drugs  networks  mobile-phones  crime 
february 2013 by jm
Network graph viz of Irish politicians and organisations on Twitter
generated by the Clique Research Cluster at UCD and DERI. 'a visualization of the unified graph representation for the users in the data, produced using Gephi and sigma.js. Users are coloured according to their community (i.e. political affiliation). The size of each node is proportional to its in-degree (i.e. number of incoming links).' sigma.js provides a really user-friendly UI to the graphs, although -- as with most current graph visualisations -- it'd be particularly nice if it was possible to 'tease out' and focus on interesting nodes, and get a pasteable URL of the result, in context. Still, the most usable graph viz I've seen in a while...
graphs  dataviz  ucd  research  ireland  twitter  networks  community  sigma.js  javascript  canvas  gephi 
january 2013 by jm
Patent trolls want $1,000 for using scanners
We are truly living in the future -- a dystopian future, but one nonetheless. A patent troll manages to obtain "gobbledigook" patents on using a scanner to scan to PDF, then attempts to shake down a bunch of small companies before eventually running into resistance, at which point it "forks" into a bunch of algorithmically-named shell companies, spammer-style, sending the same demands. Those demands in turn contain this beauty of Stockholm-syndrome-inducing prose:

'You should know also that we have had a positive response from the business community to our licensing program. As you can imagine, most businesses, upon being informed that they are infringing someone’s patent rights, are interested in operating lawfully and taking a license promptly. Many companies have responded to this licensing program in such a manner. Their doing so has allowed us to determine that a fair price for a license negotiated in good faith and without the need for court action is a payment of $900 per employee. We trust that your organization will agree to conform your behavior to respect our patent rights by negotiating a license rather than continuing to accept the benefits of our patented technology without a license. Assuming this is the case, we are prepared to make this pricing available to you.'


And here's an interesting bottom line:

The best strategy for target companies? It may be to ignore the letters, at least for now. “Ignorance, surprisingly, works,” noted Prof. Chien in an e-mail exchange with Ars.

Her study of startups targeted by patent trolls found that when confronted with a patent demand, 22 percent ignored it entirely. Compare that with the 35 percent that decided to fight back and 18 percent that folded. Ignoring the demand was the cheapest option ($3,000 on average) versus fighting in court, which was the most expensive ($870,000 on average).

Another tactic that clearly has an effect: speaking out, even when done anonymously. It hardly seems a coincidence that the Project Paperless patents were handed off to a web of generic-sounding LLCs, with demand letters signed only by “The Licensing Team,” shortly after the “Stop Project Paperless” website went up. It suggests those behind such low-level licensing campaigns aren’t proud of their behavior. And rightly so.
patents  via:fanf  networks  printing  printers  scanning  patent-trolls  project-paperless  adzpro  gosnel  faslan 
january 2013 by jm
Universal properties of mythological networks - Abstract - EPL (Europhysics Letters) - IOPscience
Abstract:

As in statistical physics, the concept of universality plays an important, albeit qualitative, role in the field of comparative mythology. Here we apply statistical mechanical tools to analyse the networks underlying three iconic mythological narratives with a view to identifying common and distinguishing quantitative features. Of the three narratives, an Anglo-Saxon and a Greek text are mostly believed by antiquarians to be partly historically based while the third, an Irish epic [jm: "An Táin Bó Cúailnge", The Tain, to be specific], is often considered to be fictional. Here we use network analysis in an attempt to discriminate real from imaginary social networks and place mythological narratives on the spectrum between them. This suggests that the perceived artificiality of the Irish narrative can be traced back to anomalous features associated with six characters. Speculating that these are amalgams of several entities or proxies, renders the plausibility of the Irish text comparable to the others from a network-theoretic point of view.


Here's what the Irish Times said:

The society in the 1st century story of the Táin Bó Cúailnge looked artificial at first analysis of the networks between 404 characters in the story. However, the researchers found the society reflected real rather than fictional networks when the weakest links to six of the characters are removed.

These six characters included Medb, Queen of Connacht; Conchobor, King of Ulster and Cúchulainn. They were "similar to superheroes of the Marvel universe" and are "too superhuman" or too well-connected to be real, researchers said. The researchers suggest that each of these superhuman characters may be an amalgam of many which became fused and exaggerated as the story was passed down orally through generations.
networks  society  the-tain  epics  history  mythology  ireland  statistics  network-analysis  papers 
july 2012 by jm

Copy this bookmark:



description:


tags: