# Topic covered
* Memory & Storage Systems
  * RAM, ROM, HDD, SSD
* Databases Types
* Database replication
  * Synchronous replication
  * Asynchronous replication
  * Single-Master Replication
  * Multi-Master Replication
* RAID - Redundant Array of Independent Disks

Memory & Storage Systems

Random access memory (RAM)

• RAM the primary internal memory of the CPU.
• Volatile memory
• Types - SRAM, DRAM

SRAM

• Static random access memory (SRAM)
• Stores data as long as there is power in the system
• It’s more expensive than DRAM
• Used for - cache memory

DRAM

• Dynamic random access memory (DRAM)
• Slower and cheaper than SRAM
• It refreshes much more frequently
• Types
  • SDRAM - Synchronous Dynamic Random Access Memory
  • DDR SDRAM - Double Data Rate Synchronous Dynamic Random-Access Memory
    • Types: DDR4, DDR5

Read-only memory (ROM)

• ROM is non-volatile and stores data permanently
• Usage: Firmware, BIOS

Hard Disk Drive(HDD)

A hard disk drive is an electro-mechanical data storage device that stores and retrieves digital data using magnetic storage with one or more rigid rapidly rotating platters coated with magnetic material. It is a non-volatile storage device, meaning it retains stored data even when powered off.

HDDs are typically installed internally in a computer and are attached directly to the disk controller of the computer’s motherboard.

Solid State Drive(SSD)

Solid-State Drive is a new generation of storage device used in computers. It replaces traditional mechanical hard disks with flash-based memory, which is significantly faster.

SSDs are more resistant to physical shock, run silently, and have higher input/output rates compared to hard disk drives. They store data in cells, and the number of bits stored in each cell determines their properties.

Other

• USB/Flash/Pen drive
• SD cards

Databases

SNC: DBMS and Databases: /categories/database/

SQL or Relational DB

• Fixed Schema - in table from
• Follow - ACID properties
  • Atomicity, Consistency, Isolation, Durability
• Easy vertical scaling, difficult horizontal scaling

NoSQL or Non-Relational or Document DB

• No fixed Schema - key-value pair
• Highly Scalable, Sharding
• Dynamic data flexibility
• Used for both - heavy read and write operations

Column DB

• Mid-way of relational and document db
• Used for - heavy writes operation
• Do not follow - ACID properties
• Eg: Cassandra

Search DB

• Eg: Elasticsearch, Solar

There are many techniques to scale a relational database: master-slave replication, master-master replication, federation, sharding, de-normalization, and SQL tuning.

Database replication

Data Replication is the process of storing data in more than one site or node. It is useful in improving the availability of data.

It is simply copying data from a database from one server to another server so that all the users can share the same data without any inconsistency.

Advantage

• Data replication allows for improved data backup
• Handle fault tolerance i.e. users can access data stored at other nodes if current fails
• Improves the performance for retrieval
• Also improves the availability of data

Disadvantage

• Replication adds more hardware and additional complexity.

Consistency Models or Consistency Algorithms

Synchronous replication

• Aka - Read after write consistency
• Master sent write issued and waits for ack from each replica

# Pros
* Replication lag is Zero
* Data is always consistence

# Cons
* Has performance hit
* Write fails - if a single replica ack is missed

Asynchronous replication

• Master sent write issued and don't wait for ack from each replica

# Pros
* High read performance

# Cons
* May have some inconsistency

Types of data replication methods

• https://github.com/donnemartin/system-design-primer#master-slave-replication
• https://github.com/karanpratapsingh/system-design#database-replication

Single-Master Replication

Aka Master-Slave replication.

The master serves reads and writes, replicating writes to one or more slaves, which serve only reads. Slaves can also replicate to additional slaves in a tree-like fashion.

If the master goes offline, the system can continue to operate in read-only mode until a slave is promoted to a master or a new master is provisioned.

Use cases

• As Data warehouse - the slave node can be used for report generation
• As CDN - place salve node close to end users

Disadvantage

• Additional logic is needed to promote a slave to a master.

Multi-Master Replication

Master-Master replication

Both masters serve reads and writes and coordinate with each other on writes. If either master goes down, the system can continue to operate with both reads and writes.

Disadvantage

• Load balancer is needed to determine where to write.
• Most master-master systems are either loosely consistent i.e. lazy and asynchronous,violating ACID)
• It may have increased write latency due to synchronization.

RAID - Redundant Array of Independent Disks

It is a technique that combines multiple physical disk drive components into one or more logical units for the purposes of data redundancy, performance improvement, or both.

RAID provides several benefits:

Increased data reliability: RAID provides redundancy, which means that if one disk fails, the data can be recovered from the remaining disks in the array. This makes RAID a reliable storage solution for critical data.

Improved performance: RAID can improve performance by spreading data across multiple disks.

RAID-0 (Stripping)

• Data are split up into blocks that get written across all the drives in the array
• Min 2 drive is required

When you save a file, RAID 0 breaks the data into segments called striped units. Then it spreads that data across all of the drives in your array. This is called striping, and it helps you access data faster because you have multiple drives working together to read, write, and store data.

# Pros
* Great performance

# Cons
* Not fault-tolerant
* No redundancy So, no backups

RAID-1 (Mirroring)

• Data are stored twice by writing them to two data drive
• Min 2 drive is required

Use case

• RAID-1 is ideal for mission-critical storage, for instance for accounting systems.
• It is also suitable for small servers in which only two data drives will be used.

# Pros
* Fault-tolerant --> Data protection

# Cons
* Storage capacity is only half
* Performance over-head writing in two drive

RAID10 or RAID 1+0

• Hybrid RAID configuration – RAID 1 + RAID 0
• Stripe the data and then mirror it
• Min 4 drive is required

If you have at least four drives, RAID 10 will increase the speed that you would have with just one drive, and you get the advantages of having redundancies. However, this also means that you have to buy more drives, and you only get half the capacity of all of them.

# Pros
* High performance
* High fault tolerance
* The rebuild time is very fast 

# Cons
* Expensive, multiple disk usage
* Capacity is only half

RAID-5 (Block-Level Stripping with one Parity)

• Striping with one parity
• 3 or more drives is required
• If 4 drive is uses
  • Data is divided into 3 blocks and stored in 3 disk
  • Parity is stored in 4th disk

When you write data in this array, just like in RAID 0, your data is broken down into units and spread over the hard drives in your array. But, aside from striping the data, it also stores parity bits on the drives. parity bit is an additional binary digit that helps your array check if there’s any error or missing segments. These bits of data can also serve as redundancies.

# Pros
* Only 1 disk is used from redundancy
* Can handle max `1 disk failure`

# Cons
* This is complex technology

RAID-6 (Block-Level Stripping with two Parity Bits)

• Striping with double parity
• 4 or more drives is required
• If 4 drive is uses
  • Data is divided into 2 blocks and stored in 2 disk
  • Parity is stored in other 2 disk

# Pros
* Can handle max `2 disk failure`

# Cons
* Write data transactions are slower than RAID 5 due to the additional parity
* This is complex technology

• What about RAID levels 2, 3, 4 and 7?
  • These levels do exist but are not that common.