If we are trying to solve optimization problems, sometimes we will end up with a mathematical programming which requires modelling the problem and solve it with some solvers. For example, when we are computing VM placement in data centers, we can use this kind of optimization to minimize usage of computer resources and network utilization. IBM ILOG CPLEX Optimizer is a powerful solver that can help us easily solve this kind of problems. CPLEX includes CPLEX studio, an interactive command line to ...

ARP is an essential protocol in today’s network which establish a connection between link layer and network layer. We all know that web address can be translated into IP address with DNS. But how can we know the hardware address of a certain IP address so that we can form the Ethernet header of a packet? ARP is our answer. From the upper image we can understand how ARP works. Basically it’s about asking “Which MAC address has this IP address?” and answering “I’m occupying this IP address with t ...

Till now, we’ve already learned how a fat tree topology can help introduce flexibility into data centers and lower the cost on high-end device. From the programming aspect, we’ve learned use mininet to simulate a virtual network and implement an OpenFlow controller with Ryu in python. However, there are still many successful studies on data center network. One of them is the famous Portland system, which may be implemented in the following articles. Besides, to introduce traffic engineering into ...

DHCPDHCP is a protocol helping assigning clients IP addresses in a Local Area Network (LAN). DHCP consists of 4 steps: DHCP discover, DHCP offer, DHCP request and DHCP ACK. DHCP discover is like client broadcasting in LAN finding a DHCP server (often located at the router) who can give it an IP address. DHCP offer is the server offering a possible IP, while DHCP request is the client broadcasting to all other clients and the server that it is going to take that IP. The last step often differs i ...

When we are implementing a fat-tree topology datacenter, we cannot use a simple learning switch to connect the network since simple switch does flooding for unknown destination thus in fat-tree there will be packets stuck in the loop. So we need to use openflow to program the controller to controll switching. In this article, we first implement a simple example where a controller controlls packet switching in a 2-linear network like the figure below. In this topology, when we do “h1 ping h2”, h ...

A simple learning switch is a switch which can learn forward messages from packets received. It learn source MAC address and input port from new packets and store messages in its own table. For packets it don’t know where to forward it simply floods them. If we replace simple learning switch with a Openflow dumb switch, the main difference will be the logic will be implemented in the controller. The flow chart will be like this. From the flow chart, we can see that there are two tables, one in ...

The servers in a data center are also organized into racks. Servers in a single rack are all connected to a Top of Rack (ToR) switch. The ToR switches in an area are connected to the same aggregation switch, and aggregation switches are all connected to core switches which link to the network outside the data center. To depict such a tree topology, we can refer to the image below: However, as we can see from the image, the higher level links need to have more bandwidth just cost more. Besides, ...

Jitsi Meet is a open-source video conference web application. It consists of 3 parts: Jitsi Meet Web app, dealing with user request and http response Jitsi Videobridge, transferring video stream data between each of users using P2P-like connection Jitsi Cofo, Jitsi video conference, the most complicated part of jitsi. It handles almost all important parts like handling XMPP connection and message, managing different videobridges, separate chat rooms and individual users. Installation of Jitsi ...

It’s convenient to construct a matrix in Python using multiplication: 123>>> mat = [[0] * 5] * 5>>> mat[[0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [0, 0, 0, 0, 0]] However, we can meet some problems when we want to modify only one element. 123>>> mat[0][0] = 1>>> mat[[1, 0, 0, 0, 0], [1, 0, 0, 0, 0], [1, 0, 0, 0, 0], [1, 0, 0, 0, 0], [1, 0, 0, 0, 0]] We can found that by modifying only one element of the matrix, we modify the whole col ...

Welcome to Hexo! This is your very first post. Check documentation for more info. If you get any problems when using Hexo, you can find the answer in troubleshooting or you can ask me on GitHub. Quick StartCreate a new post1$ hexo new "My New Post" More info: Writing Run server1$ hexo server More info: Server Generate static files1$ hexo generate More info: Generating Deploy to remote sites1$ hexo deploy More info: Deployment