Introduction

Purpose: Introduce participants to network and computing architectures to enable effective scientific computing.

Background: Scientific synthesis fundamentally depends on the use of computers and the Internet. To complement the practical understanding of how networks of computers are structured, this lesson reviews basics of computer networks and server-based computing in the service of scientific synthesis. In particular, the lesson reviews how computers are connected in networks, the relationship between networks, computer addresses, and web site domain names, and the utility of remote computing and computing in the ‘Cloud’ for speeding up scientific analysis.

Learning outcomes

  1. To understand the basic architecture of computing networks
  2. To conceptualize how computer hosts relate to web sites
  3. To learn to connect to a remote computer via a shell
  4. To appreciate the value of remote and cloud computing

Background

The Internet

Our goal is to:

  • Provide basic background on computer networking
  • Lay groundwork for understanding tools that use the Internet

Basic networking

  • Host computers connect via networking equipment
  • Can send messages to each other over communication protocols
  • Client: the host initiating the request
  • Server: the host responding to a request

Local networks

  • All messages broadcast on the local network
    • Local hosts only “listen” for data that targets them

Routed networks

  • Mechanism to connect networks together via forwarding data along routing paths

IP Numbers and Hostnames

  • Hosts are assigned a unique address used for all communication and routing
    • Internet Protocol Address (IP Address)
      • 128.111.220.7
  • Each IP Address can be used to communicate over various “ports”
    • Allows multiple applications to communicate with a host without mixing up traffic
  • IP numbers can be difficult to remember, so also assign hostnames
    • Handled through the global Domain Name System (DNS)
    • Clients first look up a hostname in DNS to find the IP Number
    • aurora.nceas.ucsb.edu -> 128.111.220.46
    • Then they open a connection to the IP Number

Client-Server and the Internet

Peer to Peer networking

Internet versus the Web

What is the difference?

  • Internet: Global communication network using the Internet Protocol (IP)
    • Supports many application protocols, including:
      • HTTP (web)
      • FTP (file transfer)
      • SMTP (email)
      • SSH (secure shell)
  • Web: Global system of interlinked hypertext documents
    • Transported over HTTP application protocol (over IP)
    • Generally accessed using web browsers that retrieve the documents

Routing on the internet

  • 80 routers at the core
  • dense “mantle” of routers at the periphery
    • loss of the core still leaves > 70% of internet connected peer-to-peer

Internet Application Protocols

  • Web (http)
  • Email (smtp)
  • Secure Shell (ssh)
  • Telnet (telnet)
  • File Transfer Protocol (ftp)
  • Network News (nntp)
  • …

The Web

  • Global network of documents hosted on web servers around the world
  • Web documents connected using hyperlinked HTML documents
  • Other document types common (images, other media)

Secure Shell

  • Used for connecting to and running shell commands on a remote host
  • Tremendously simplifies remote computing
  • Supported out-of-the-box on Linux and Macs

Exercise

Launch your Terminal program

  • MacOS: Terminal Navigate to Applications | Utilities and open Terminal
  • Windows: Git Bash Navigate to Windows Start | Git and open Git Bash
    • Note: Alternate and possibly better terminal for Windows is cmder

Connecting to a remote server

jones@powder:~$ ssh jones@aurora.nceas.ucsb.edu
jones@aurora.nceas.ucsb.edu's password: 
jones@aurora:~$

Change your password

jones@aurora:~$ passwd
Changing password for jones.
(current) UNIX password: 
Enter new UNIX password: 
Retype new UNIX password:
  • Add your new password to your password manager!
  • Later you will need it to connect over the web to RStudio Server

Modern cloud computing

In this session, we are introducing you to computing on a cluster, which is just another name for a remote computer with a lot of processors. However, computing is evolving rapidly, and there are many new technologies emerging to help with scientific computing, and to speed it up. The maturing field of cloud computing is certainly useful to researchers conducting synthesis because it allows them to access really large computing systems without having to purchase the hardware. Amazon, Google, and Microsoft are among the many companies trying to capture the demand for computing infrastructure through cloud platforms. In each, the cloud computing environment enables users to create a virtual machine, which looks and acts like a standalong server, but in actuality is sharing an underlying host computer with many other virtual machines.

Cloud computing is also evoloving mechanisms to make those virtualized platforms easier to create and maintain with the advent of container computing systems. Systems like Docker for building virtual containers are beyond our scope, but further reading is available online:

Preethi Kasireddy. A Beginner-Friendly Introduction to Containers, VMs and Docker.