Chapter 13 - Computer Becomes A Network PDF

Chapter 13 - Computer Becomes A Network Subjects History of Science The modern web bridges the different modes of computing How did the Web solve the cons of mainframes? Before the introduction of the Web, PCs did n...

Chapter 13 - Computer Becomes A Network Subjects History of Science The modern web bridges the different modes of computing How did the Web solve the cons of mainframes? Before the introduction of the Web, PCs did not have internet access. Cons : expensive, annoying to use and hence needed trained professionals. The web allowed (untrained) personal computer users to do almost everything a mainframe allowed but cheaper and from their homes. e.g. people could register from their own PCs, reducing call center employees. Productivity paradox : in the 70s, computers were clearly meant to benefit businesses, but the economic growth was slow (contrasting). 90s → 00s : economic boom, thanks to computers. Data Centers Massive increase in server importance in the 90s. During 90s, focus of firms were on making their servers more powerful. After the introduction of the Internet, many large websites starting building these servers as traffic increased. Some of them had built server farms. Chapter 13 - Computer Becomes A Network 1  Mainframe servers PC based servers Why did Itanium not work? Itanium was supposed to be a very powerful processor. Too expensive to integrate. The market had shifted. Datacenters were now focusing on cheaper alternatives like combining computer boards (first notably done by Google). Chapter 13 - Computer Becomes A Network 2  This cheaper alternative was also easy to expand - just combine more computer boards. Intel’s previous processors provided good enough performance for cheaper prices. Cloud Computing Until now, companies were running their own complex, expensive datacenters. Google was created for web searching and web advertising. This of course meant handling of massive amounts of data (aka more data centers right?) They worked around the conventional expensive data servers approach by using cheap computer boards. For cooling, they used water instead of air- conditioning. But their biggest contributions came from MapReduce - an efficient algorithm that splits tasks into multiple processors which work on them in parallel, and then merging the results. Using this, they were able to cheaply distribute several webpages for instance, and find key words, then return us the results. Was this reliable? Not as much as the conventional, expensive servers. However, Google File System, an additional layer of software, made sure that data was stored redundantly, making it seem as though disk drives never failed (they failed a lot). Birth of cloud services Thanks to Google, cloud services took off, offering companies to take away all the server hassles for cheap. e.g. Despite Netflix requiring large server capacity, it hosts its systems on AWS (cloud service), and does not have its own data center. Processors Why was higher clock cycles preferred between 90s and 00s? It was an indication of good performance. Most of the programs back then were single threaded. Chapter 13 - Computer Becomes A Network 3  Single threaded programs performed better with higher clock cycles. What caused the change in mindset towards high clock cycles? Performance was getting hampered with higher clock cycles. It was getting less work done per clock cycle. Consuming more energy. This was given much importance as laptops became more popular. Intel ever since has not mentioned how many cores their processors contain, making each product having their pros and cons. PCs were not required to be changed frequently as the intel chips were so good that a decade old PC could run windows reasonably well. NVIDIA GPU - many processor cores, parallel processing - scientific computing, deep learning, gaming Mainframes - lesser but much more powerful cores - handling massive datasets, IO, security e.g. airlines. Virtualization 70s : CP/CMS First built as a solution to timesharing in TSS. Mainframe had terminals connected, and each user could have the illusion of having their separate processors. In reality, it was one physical processor that switched between each user so fast it seemed they were using their own processors. This was an idea taken from virtual memory that existed in the 60s. Not only did each terminal have virtual processors, they also could run operating systems like OS/360, CMS. How did virtualization give IBM a competitive edge? They could run multiple applications during the day (virtualization), and batch processing during the night (no virtualization). CP/CMS also had backward compatibility as it could run older operating systems. Chapter 13 - Computer Becomes A Network 4  VMs removed the need for having servers for each department. Instead, they can share one big server, and their respective software can be run on VMs. HR crashing does not affect Accounts. Video Streaming In early days, video streaming was very difficult - even after high compression, it could not be streamed in HD. Thanks to higher speed internet connections, streaming quality increased. YouTube As YouTube became popular, Google acquired it (for a massive sum of money at the time) as Google’s infrastructure could handle huge traffic. Google started monetizing YouTube users, thus further attracting users. Netflix Started off as mail-in DVD rental. Changed to streaming service after realizing the market exists, thanks to YouTube. Viewers were attracted since they paid very little and could watch at any time. Quality was bad initially but got better over time. Created its own shows/movies as they did not get licenses from production companies. Social Media Early 2000s. Web 2.0 : Part of the web that focused on the social aspects of users. e.g. Wikipedia. Websites were more dynamic and appealed to the users. Success of blog websites led to creation of Friendster, MySpace and Tumblr. These newer and more popular websites focused on the social aspects of users through media (instead of just writing). Facebook Killed every other competitor. Chapter 13 - Computer Becomes A Network 5  Started for university students since they were more sociable and keen to try new things/apps. Use grew massively. Used cheap servers like Google. Algorithm made sure users were updated about trends far more than peer updates (tactic used to beat Twitter). Farmville , likes and chat functionalities attracted users. Facebook became a all-in-one place, and profited largely from advertisements. For companies, posting ads on Facebook was common- sense. Network Computer Why did p-code not take off in the 80s? Despite allowing portability (like Java), it did not catch on as IBM PC took off. This meant that applications were being built for the IBM PC’s hardware, and not for every hardware. This allowed using full capabilities of the IBM PC. Java’s threat to Microsoft Microsoft Windows and Microsoft computers were dominant in the 90s. Then came Java, a programming language that allowed for interactive web pages / applications. The biggest pro of Java was that it’s applications could be run on any device/OS, as long as Java Virtual Machine existed. As a result, Microsoft’s applications like Word was threatened as users could just process word on web browsers or Java applications without having Microsoft PC. Why did this threat fail? Even though Network Computers had clear advantages over PCs on paper, reality was different. Internet was too slow to download everything. Chapter 13 - Computer Becomes A Network 6  JVM could not print. Java apps on the internet were agonizingly slow compared to Microsoft apps that were pre-installed. Hackers could hack into the security flaws of Sun’s Java platform. What two sectors was Java most successful in? As an OOFP for students to learn - easier to learn and debug (due to JVM). As servlets - writing into web pages and DBMS. SaaS Software as a Service. Companies subscribed to SaaS so that they did not have to maintain software themselves. e.g. Netflix does not have to build and maintain datacenters as they host it on AWS. Amazon does it for them. JavaScript and Ajax Until mid-90s, websites were static (HTML) or had some applets (Java) that were very slow and required plug-ins. 1995 : JavaScript - language that allowed coding into websites e.g. you could make sure the date input was in the correct format. Combining JavaScript with XML allowed websites to not reload when a functionality was to be accessed (Google refresh mail). This approach was dubbed Ajax. Allowed for creation of Google Docs. Welcome back, Network Computers Google’s ajax-based applications like Gmail, Google Maps and Docs were becoming more powerful. However, they needed browsers to run Ajax quickly and reliably. Chapter 13 - Computer Becomes A Network 7  They came up with Google Chrome - minimalist looking browser that ran smoothly and reliably, perfect for their applications. Chromebooks Around the time Chrome had overtaken Internet Explorer (2012), it introduced Chromebooks - a reincarnation of the Network Computers. Cheap, easy-to-manage. Using Google applications while connected to the internet - perfect for schools. Chapter 13 - Computer Becomes A Network 8