Basic Concepts

Introduction of software application named Mini Mall

Mini Mall is a simple solution for local merchants to manage there inventory online, accept online payments and generate invoice for there customers.

using this data mini mall helps merchant in book keeping, sales insights and notify about old stock to clear.

Initial Setup

so initially we started with simple setup which consists of :

• Virtual Machine / Server: A server is nothing but a computer machine designed for compute, store, and manage data, devices, and systems over a network.
• Backend: consists of API to register merchant, manage inventory, create bill, get sales insights and many more.
• Frontend: this piece mainly consists of static files such as HTML, CSS, JS to serve User Interface.
• Database: database consist of data to store merchant data, inventory data, bills data and sales transactions data.
• Web-server: software component that delivers static data like images, files, and text in response to client requests we are using Nginx here.

Loading...

Situation

So far every thing is good, let's understand the situation

• when user opens the web application using public_ip: 170.187.254.215 then NGINX as web server will serve the static files.

• we also have mobile applications for merchant which will notify the merchants about store such as payment status etc.

• for mobile applications we are offering Rest API's with which data is rendered on mobile-app. Sample API request

  curl -X GET 170.187.254.215:5000/api/inventory

Issue

Now let's understand the issue with current setup

• High Latency: Data transfer between clients and servers can be slow, negatively impacting user experience.there is no caching mechanism.

• Security Vulnerabilities: Servers may be directly exposed to the internet, making them susceptible to security attacks and breaches.

• Uneven Server Load: So tomorrow we can deploy more backend replicas which can run on different ports then we need some mechanism to balance load between all replicas.

• Bad url structure: instead of this 170.187.254.215:5000/api/user we need 170.187.254.215/api/user , so port should not be there.

Task

Hence we need both our Backend and frontend should be serve on default port: 80 .

Backend: 170.187.254.215/api/user

Frontend: 170.187.254.215/

Action

Hence here comes the first system design concept named Reverse Proxy .

Concept: Reverse Proxy

so after adding reverse proxy layer our backend is now behind the proxy server in this way we have make sure that all the traffic will went through the proxy layer and will provide some benefits as follows:

1. Url Routing : so now we don't have to mention the port in the url instead we will just add the prefix /api which in turn tell reverse proxy to hit the backend internally.
2. Rate Limiting: just by adding simple configuration we can mention the threshold to limit the request to the server.
3. Caching: reverse proxy provides the caching mechanism and we can apply this just using some configuration including proxy_path: where to store,key_zone: tells about storage,cache_key: how to create cache key etc.for more info read here.
4. Load Balancing: so in future we can add multiple replicas so one replica runs at 4000, second runs at 5000 etc.so by mentioning the list of ports reverse proxy can balance the load between different replicas.

Loading...

Situation

So far every thing is good, let's understand the situation

• we hit the endpoint 170.187.254.215 for static content and to hit the backend API's we use enpoint: 170.187.254.215/api

Issue

Now let's understand the issue with current setup

• first issue is this IP is not user friendly we need specific name to use the application.

• second issue is what if our virtual machine goes down and after starting new machine our public IP is changed.so we need a particular name to use application and it should point to the rigth IP under the hood.

Task

Hence we need a specific name like minimall.com to use the application.

Action

Hence here comes the next system design concept named Domain Name System .

Concept: DNS(Domain Name System)

At this step we have purchased the domain and in Domain management dashboard we will add a record to map our public IP with domain name.

Record Type	Name	Value	TTL (Time-to-Live)
A	minimall.com	170.187.254.215	3600

1. Record Type: A (Address record), which maps domain names to IPv4 addresses.
2. Name: The domain name is "minimall.com."
3. Value: The associated IP address is 170.187.254.215.
4. TTL (Time-to-Live): The information can be cached by DNS resolvers for up to 3600 seconds (1 hour) before needing to refresh.

Some Popular record types:

1. A (Address): Maps a domain name to an IPv4 address.
2. AAAA (IPv6 Address): Maps a domain name to an IPv6 address.
3. CNAME (Canonical Name): Alias of one domain to another (used for subdomains).

Loading...

Situation

So far every thing is good, let's understand the situation

• so now we have nice domain and our sales team, onboarded some merchants and some of the merchants are actively managing there inventory and created some bills.

• now there is some users who is using our application.

Issue

Now let's understand the issue with current setup

• for now there is no major issue but we need some visibility to get the performance metrics of the application.

• This will assist in identifying the problem before users voice their complaints.

Task

Hence we need a dashboard and a proper system to record the metrics for better visibility.

Action

Hence here comes the next system design concept named Logging and Monitoring .

Concept: Logging and Monitoring

At this step we need a system setup which will record:

1. Logs: Produced by application. Types of Logs based on Level: info, debug, error,warning etc.
2. Metrics: Mainly tells about performance of Application,Virtual Machine and Database.main performace indicators are: CPU,Memory Utilization, Latency and Network metrics.

Next thing we need a dashboard which mainly pull the record of logs and metrics from a datasource/database so that team can able to visualize and take actions.

Hence we can use Grafana or Kibana as our dashboarding tool.here we are using Grafana.

Loading...

Situation

So far every thing is good, let's understand the situation

• so now we have a nice dashboard where we can see metrics such as CPU usage, RAM usage, Disk space etc.

• we can also see the logs from dashboard which can helps us to see the Error rates,Application health,Application behaviour etc.

Issue

Now let's understand the current issue which we have observed.

• first thing which we observed is High CPU Utilization where only 20% is idle CPU.

• second thing is memory usage of virtual machine is constantly increasing.

Task

Hence we need to add some resources mainly in terms of CPU and Memory and to do that we need to create a new instance which provide more Virtual CPU's and RAM(Memory).

Action

Hence here comes the next system design concept named

Vertical Scaling