Performance of a Website

Understanding Performance in Modern Application Development

Introduction to Web Server Performance

• The course begins with an overview of modern application development, emphasizing the importance of web server performance.

• Initial considerations about performance are introduced, highlighting that while early stages of app development may not require deep focus on these issues, scalability becomes crucial as user demand increases.

Key Concepts: Latency and Its Impact

• Latency is defined as the time taken for a server to respond to a request. Understanding this metric is essential for optimizing application performance.

• Despite our perception of computers being fast, latency can be significantly affected by the speed of light, which imposes physical limits on data transmission times.

Speed of Light and Data Transmission

• The speed of light in vacuum is approximately 3 x 10^8 meters per second; however, it slows down in cables (around 2 x 10^8 m/s).

• For practical understanding, it takes about 5 nanoseconds for light to travel one meter. This translates into longer response times over significant distances (e.g., 5 milliseconds for 1,000 kilometers).

Real-world Implications of Latency

• A round trip request from Chennai to Delhi (approximately 2,000 km) results in a total latency of around 20 milliseconds if the server responds instantaneously.

• Continuous requests limited by latency mean that only about 50 requests per second can be processed under these conditions.

Bandwidth Limitations and Server Responses

• Even though latency might seem manageable at first glance, it can become problematic when servers need to handle multiple simultaneous requests.

• Each server response includes various headers and content types (HTML, JavaScript), which contribute to overall data size and processing time.

Network Capacity Considerations

• With a network connection speed estimated at around 100 megabits per second (roughly translating to about 10 megabytes per second), each kilobyte response limits throughput significantly.

• Under optimal conditions with a single server handling responses at this bandwidth limit leads to a maximum capacity of approximately 10,000 requests per second.

Scalability Challenges During High Demand Events

• In scenarios like entrance exam results being declared simultaneously across many users connecting to one server could lead to crashes due to overwhelming traffic exceeding network capabilities.

Example Response Analysis

Understanding Data Transfer and Server Scaling in Modern Applications

The Scale of Data Transfer for Google Search

• Google’s homepage may appear simple, but it transfers over 144 kilobytes of data per request due to various elements like tracking information and JavaScript.

• The complexity of the page includes dynamic modifications based on user behavior, leading to an average transfer size of around 100 kilobytes per request.

• Google processes approximately 60,000 requests per second, which translates to a potential bandwidth requirement of about 80 gigabits per second if each request requires significant data.

• This high demand necessitates distributed server architecture since a single server or even a single data center cannot handle such bandwidth requirements effectively.

• To manage this scale efficiently, systems must be designed with scalability in mind from the outset.

Memory Management Challenges in Streaming Services

• A Python HTTP server example shows that each process consumes around 6 megabytes of RAM; this is manageable for small-scale applications but problematic at larger scales.

• Long-running connections, as seen with services like YouTube during peak events (e.g., presidential debates), can lead to substantial memory consumption across many concurrent users.

• For instance, with 2 million viewers requiring 6 megabytes each, total memory usage could reach up to 12 terabytes—far exceeding typical server capacities.

• Instead of relying on one massive machine, distributing the load across thousands of servers allows for more efficient resource management (e.g., only needing 12 GB per server).

• These considerations highlight the importance of planning for scalability and resource allocation when designing applications.

Key Considerations for Application Design

• Developers should consider whether large amounts of data are necessary for every request or if smaller initial responses could suffice before scaling up based on user needs.

• Efficient use of network bandwidth and thoughtful sizing of servers are critical factors that need attention during the design phase.

What is the Size of Google's Database?

Understanding Google's Data Storage

• Google’s database size is estimated to be around 100 petabytes, which translates to 100 thousand terabytes or 100 million gigabytes, highlighting the vast scale of data managed by Google.

• Storing such a massive amount of information cannot rely solely on traditional disks; for instance, using 1 terabyte disks would require an impractical number (over 100,000) to store all this data.

The Web as a Platform

• The web serves as a versatile platform for app development, being operating system agnostic and built on robust foundations like HTTP for transport.

• At its core, a web server's function is straightforward: it receives requests and sends back information. However, creating a functional web server that meets specific needs can be complex.

Challenges in Scaling Web Servers