Congestion-avoidance tools are complementary to, and dependent upon, queuing algorithms. Queuing/scheduling algorithms manage the front of a queue, while congestion-avoidance mechanisms manage the tail of a queue.
Congestion-avoidance tools are designed for TCP traffic, because TCP has built-in flow-control mechanisms that operate by gradually increasing traffic flows until packet loss has occurred. Once packet loss has occurred, the transmission rate is reduced before slowly ramping up again. This means that if no mechanism is in place to control TCP, any particular flow has the ability to eat up all available bandwidth.
When there are no congestion-avoidance tools in place, and queues fill, tail drop occurs, which means all traffic is dropped.
In a constricted channel without congestion-avoidance tools, TCP connections eventually synchronize with each other – they ramp up together, lose packets together, and back off together. This is called global synchronization and basically results in “waves” of TCP traffic.
Congestion-avoidance tools has no real benefit or use for UDP traffic, because UDP traffic does not have any retry logic.
RED combats global synchronization by preemptively and randomly dropping packets before queues fill. Instead of waiting for the queues to fill, Continue reading
In Malcolm Gladwell’s book The Tipping Point [affiliate link] the author identifies key roles that individuals play in spreading knowledge and ideas. Gladwell outlines two key roles, the maven and the connector. The maven is a person who accumulates subject matter expertise and is willing to distribute that knowledge on request. Think of a maven […]
The post The value of Connectors in the workplace appeared first on Packet Pushers Podcast and was written by John Harrington.
I am selling my rack as I need to make some space and I am now using CCIE Rack Rentals all the time which I find are so much more convenient. Only for UK as it is collection only, but if you are in the market for a CCIE Rack based on the INE topology […]
The post CCIE Rack for Sale appeared first on Roger Perkin - Networking Articles.
I recently decided to commit to my current company. I don’t mean work harder or focus more, though that’s implicit. I mean to consider being a “lifer.” For an independent personality like myself that’s a pretty tall leap. One of the reasons I got my CCIE was to avoid ever being the guy sweating when […]
The post Ageism and your career in I.T. appeared first on Packet Pushers Podcast and was written by Keith Tokash.
As some of you may know, I have spent a fair amount of my time in the last few years designing and improving multi-tennant hosting environments. Each revision attempts to learn from the mistakes of the previous iterations, as well as bundle in new features and “advancements” from each of the different vendors in the stack.
New offerings on the storage fronts, developments in the server space in the form of the boom of virtualisation, and the simple existence of the network amongst the fact that none of these technologies changed the existing/fundamental laws of networking.
Software-Defined Networking has sprung up as a way of providing both advancements in our current architectures and providing agility in changes needed in the future, but what is truly needed is a true abstraction of the entire data centre model that encompassed all of compute, storage, security and networking. The ability to define all of the requirements of your existing data centre and have them deployed and rolled out across which ever stack you are using (Private / Public / Hybrid / Tomorrows Favourite buzz.), in a consistent and definable manner.
Merging the requirements of each of the existing silos and describing Continue reading
As a systems admin a few years back I decided that I needed to figure out a way to understand networking. I could do basic things, but how do you figure out what you don’t know? That’s when I started listening to podcasts. The first podcast I came across was Packet Pushers and I haven’t […]
The post AdaptingIT: Why I Started a Podcast Featuring Women in Tech appeared first on Packet Pushers Podcast and was written by Lauren Malhoit.
A lot of us struggle in our careers when it comes to accepting or declining work, whether it be voluntary or voluntoldary. Especially the individuals in our field that strive to succeed. We see it as another opportunity to stand up to the plate and knock one out of the park. The problem, however, is that […]
The post When it’s OK to say “No” appeared first on Packet Pushers Podcast and was written by Edward Henry.
If you’ve been following Embrane over the past several months, you know we’ve been focusing almost exclusively on differentiating our business in the SDN space by promoting the fact that we have been the only company securing and announcing a steady stream of paying, in-production customers. As a result, we’ve been placing less emphasis on touting the advancements we’ve made on the technology side. However, since it’s our technology leadership that’s attracting our rapidly growing customer base, it’s time to show off our technology chops too. This week our SDN leadership becomes even more evident with the introduction of our new application-centric networking solutions.
(Okay, it’s not exclusively a technology announcement since we announced another new customer, Ryan Labs Asset Management. More on them in a bit.)
While most vendors started their SDN movement from the bottom up, looking at ways to add agility at the connectivity layer (a.k.a. Layer 2), Embrane continues to take a top-down approach to the network. We focus on the network services that support, enhance and secure the ever-growing number of business applications in an enterprise data center. After all, applications drive enterprises business. The newest release of the Embrane heleos Continue reading
How does the internet work - We know what is networking
I can say at the beginning that more and more fellows is visiting this blog. Finally networking knowledge project did succeed to get some fans from out there so I felt the need to thank you all for visiting this place. As you can see here from the blog post list I didn’t follow this movement myself these […]
As a follow-on to my previous article on onePK – Cisco onePK: Now I Get It – I recorded a screencast in which I talk about what a onePK-enabled network is capable of. I also demonstrate two applications which make use of onePK to gather telemetry from the network and also program the network.
Please leave a comment below with any questions or feedback.
In this community show recorded at Cisco Live US 2013 in Orlando, Cooper Lees, Lauren Mahoit and Colin McNamara join Packet Pushers hosts Greg Ferro and Ethan Banks in a happy chat about the coming world of networking automation. We discuss the following high-level topics. Automation, orchestration and devops in the network is becoming an ever-bigger deal. We apply this in a Cisco […]
The post Show 154 – A Bit of Swamp Ass to Lift Your Day – CLUS 2013 Community Show appeared first on Packet Pushers Podcast and was written by Ethan Banks.
At Cisco Live US 2013 in Orlando, Packet Pushers co-host Ethan Banks was joined by CCDE program manager Elaine Lopes, CiscoPress author Russ White who was closely involved with the CCDE program creation, quad-CCIE & CCDE Scott Morris, and CCIE & CCDE Jeremy Filliben who instructs a CCDE bootcamp. We discuss (what else) how to […]
The post PQ Show 29 – How To Prepare for the CCDE appeared first on Packet Pushers Podcast and was written by Ethan Banks.
Of all the tools within the QoS toolset, congestion management tools, also known as queuing tools, provide the biggest impact on application service levels. Whenever packets enter a device faster than can exit it, congestion exists and this is where queuing tools come into play. Queuing tools are only engaged when congestion exists, otherwise packets are sent as soon as they arrive. When congestion does exist, packets must be buffered, or queued, to mitigate dropping.
Packet markings, or lack thereof, affect queuing policies, so queuing policies are complementary and have a dependence on classification and marking policies.
These two terms are often incorrectly used interchangeably – they are two different things. Scheduling determines how a frame or packet exits a device. Whenever packets enter a device faster than they can exit it, as is the case with speed mismatches (ex. Gigabit Ethernet traffic heading to a WAN interface), congestion can occur. Devices have buffers that allow the temporary storing and subsequent scheduling of these backed-up packets, and this process is called queuing.
Inbound traffic > Queuing (During congestion) > Scheduling > Outbound traffic