What is Scalability? (Thu Jan 20, lect 2) | previous | next | slides |

What do we mean by Scalability?

• This course is about Scalability but what do we mean by that?
• From a dictionary: “the capacity to be changed in size or scale”
• This could be organizational, operational or in our case, technical

What do we mean by scaling?

• Your program works not just with 100K of data but with 100Gig of data
• Excel can handle a model not just with 300 rows but with 300K rows
• Your web site works when 10 friends are using it but fails miserably when 5,000 people are using it simultaneously

There’s a life cycle

• Sin: “Premature Optimization”
• Don’t assume that you know why
• You are probably wrong!
• Always measure to determine where the bottleneck is
• You will never find the “last” bottleneck.
• You will remove bottlenecks until the system scales to your requirements
• “You’re not Google” (or facebook, or instagram, or …)

Lets form some teams

• How is it going with Ruby so far?
• Let see if we can form some teams of 3 students… how?
• Form 3 random Zoom Breakouts. Get to know each other
• Each breakout will go for 5 minutes and then we do it again

Software Aspects

• You need to start with an idea of your goal
• Usually expressed as:
  1. Simultaneous users
  2. Performance under load
  3. Expensive operations
  4. Rate of operations
• Needs to have a target range that can be measured
• Search for bottlenecks
• Engineering for scalability is not free
• Tradeoff with cost (people or equipment)
• There are Best practices

Hardware Aspects

• Scaling “up” versus “out”
• Have an intuition about the speed and latency of different media
• What’s the difference?
• Compare speeds of:
  1. memory
  2. Local disk
  3. SSD Solid State disks
  4. NAS Network Attached Storage
  5. Local LAN communications
  6. Internet communications

Operations aspects

• Similarly there are goals
• Usually includes
  1. Availability
  2. Freshness
  3. Support
• Availability sometimes comes as “Terms of Service”
• Reliability of the infrastructure and quality of the software
• Timely response to temporary load variability

Development Process aspects

• Embrace change because it will happen
• Agile methodologies
• Automated testing
• Continuous integration, testing and deployment
• Design for maintainability

Thank you. Questions?  (random Image from picsum.photos)