I was overloaded during the last few weekends and my Inbox is yet again overflowing with links to excellent content. For a warm-up, look at the eight levels of vendor acceptance (a side effect of a really tough lab test during the EuroNOG 2011 conference).
On a more serious note, the most useful article of this week is probably the BGPmon Web Services API that describes how you can query the global BGP table through whois or SOAP.
Data Center and virtualization
Chris Marget shares some very interesting IGMP-related vSwitch behavior. The number of kludges they implemented to avoid deploying standard bridging code is astounding.
BFD seems to be implemented on physical ports on the Nexus 7000, so it works great on P2P L3 interfaces, but not so well on VLAN interfaces.
The real news is not that Facebook servers 1 trillion pages per month ... but how many requests they have to process internally to generate that content.
Jonathan Hudson wrote an interesting article comparing the current TRILL/SPB debate to the old IS-IS/OSPF religious wars. He’s right, but he just might be defending an evolutionary dead end. With MAC-over-IP (or IP-over-IP) virtual networking (VXLAN and NVGRE) being introduced in enterprise data centers, the need for large-scale bridging is gone.
Automation conundrum – a good look at what we actually need to implement blah blah cloud infrastructure. Bonus points for introducing cloudwashing.
Amazon, Netflix, Standard Cloud APIs and the Inevitable Lock-in. Until we get a standard cloud API (not likely for at least a few years), we’ll face the same problem we’ve been facing with databases (or networks) for the last 30 years – whenever you want to use an interesting feature, you have to tie your implementation to a specific vendor’s products.
5 Scalability Poisons And 3 Cloud Scalability Antidotes – another great scalability 101 recipe. If you want to outgrow the “mission-critical” craplications, start reading the High Scalability blog – it has plenty of excellent links to stuff that really matters.
Another scalability success story: RightScale has launched its 3 millionth server. In totally unrelated news, some people are forced to migrate running VMs between data centers to implement disaster avoidance for “mission critical” application.
Paul Gear nicely exposed many things we take for granted in the IPv4 world that will be gone in NAT-less IPv6 world. Yeah, it will be an interesting journey.
Routing IPv6 in 2011 – looks like the IPv6 BGP table explosion won’t be as bad as I thought it would be.
Take the Facebook challenge – point facebook.com to 127.0.0.1 in your /etc/hosts, start a local web server (for logging purposes only) and watch in horror all the referrers in the server logs.
Changing bugs into fads (Functions-as-Designed) – if you have too many problems, you simply declare they are not bugs but design features. Great job, DHS.
Application Analysis Using TCP Retransmissions, Part 1 ... interesting. Part 2 (buffering) should be even better.