Cloud Classroom Series – 25/Jul/2020

TCP/OSI Model

• This model uses layers to help give a visual description of what is going on with a particular networking system
• OSI Model
• Each Layer has funcationalities
  • Layer 1: Physical: This layer defines the logic levels, data rate, physical media & data conversion functions
  • Layer 2: Data link: Package & Unpackage the data in frames
  • Layer 3: Network: This network layer handles packet routing and switching
  • Layer 4: transport: This layer provides the quality of service functions. TCP is part
  • Layer 5: Session: This layer handles authentications & authorization
  • Layer 6: Presentation Layer: This layer checks the data & it can handle encryptions & decryptions
  • Layer 7: This layer used communication protocols like http, IMAP, SMTP.
• When we speak of cloud we use tcp/http(s), So we need to worry about Layer 4 & layer 7
• Transport Layer Protocols
  • TCP
  • UDP
• Port is defined at the Transport Layer (Layer 4)

Multiple Servers to Handle Load

• As the load increase on web/application servers we can solve the issue

  • By increasing RAM/CPU (Vertical Scaling)
  • By increasing servers and sharing load b/w them (Horizontal scaling)

• Most of the organizations prefer Horizontal scaling especially in cloud

• Since the multiple servers are involved, we would add a load balancer and use that as an interface to our multiple servers

• There are two types of load balancers

  • Software Load balancers
  • Hardware Load balancers

• Load balancer whenever a client sends a traffic to which server should i forward?

• To solve this load balancers have load balancing algorithms

  • Round-Robin: Requests are evenly distributed across servers sequentially
  • Least Connections: A new request is sent to the server which has fewest connections at that moment
  • Least time: Sends the request to server which has the fastest response time
  • Sticky Sessions: Sends the request to the same server that client was given last time
  • Hash: Define a key such as client IP address or request url to send to a particular server
  • Random:

• Load balancing can happen at two layers

  • Layer 4
  • Layer 7

• In Layer 4 loadbalancing, Load balancing can be done by redirecting the traffic recieved on a particular port of loadbalancer (source port) to particular port of connected servers (target port)

• Layer 7 loadbalancing

Domain Names and DNS Servers

• Generally we access the serves with names. There are two kinds of names depending on purpose

  • public: https://google.com, https://swiggy.com
  • private: https://qt.internal.com

• DNS is used to give names. It is like the phonebook (Telephone directory) of internet.

• When i type google.com there will be some server which is hosting google.com and all communication protocols commmunicate mostly using ipaddresses not names.

• So when i type name how does it gets converted to an IP Address, For this DNS Servers are maintained.

• DNS server will store

  • records mapping name to ip address

• Generally organization for their internal names use internal dns servers and websites use public dns servers.

• To store these Mapping DNS servers have different record types

  • A record: This record holds the ip address of domain
  • CNAME Record: Forwards one domain name to other

  fb.com
  

  • MX record: Directs mail to email server
  • TXT record: lets admin store text in the records
  • NS Record: Stores the name server for DNS entry

• If it is private names for your network purpose we need to maintain dns entries

• If it is for public website, we have Domain Name Sellers and hosting services like GoDaddy, AWS, Azure