System design

what Cassandra is

wide-column store

Couchdb

alternative to MongoDb. An open-source document-oriented NoSQL database, implemented in Erlang

four examples of Wide-column store

Apache Cassandra, Apache HBase, BigTable, scylla

Database replication master/slave

Master -original, only writes, slaves - copies, only reads

expand CDN

Content Delivery Network

expand SPOF

Single point of failure

Cache eviction policies

LRU - Least recently used, LFU - Least Frequently used, FIFO - First in First out

CDN - describe what it is

A CDN is a network of geographically dispersed servers used to deliver static content. CDN servers cache static content like images, videos, CSS, JavaScript files, etc.

Data Lake

is a system or repository of data stored in its natural/raw format,[1] usually object blobs or files

Stateless Web Tier

session is kept in NoSQL database. It is no longer kept in web server. Each server in the cluster can access state data from the database

Load balancer, sticky session

user who authenticated in one of the servers in a cluster, will be redirected to the same server for future calls (for certain amount of time that is longer than the session timeout)

Disadvantages of sticky session

more difficult to add or remove servers, more difficult to handle server failures

Advantages of stateless web tier vs stateful web tier

simpler, more robust and scalable

Shared data store for state data (sessions), what to use

memcached/Redis, NoSQL

geoDNS

geolocation DNS routing - to make sure, that the client is directed to the closest datacenter

Logging and monitoring tool

Datadog, Matomo, prometeus, Pendo

example of Key business metrics

daily active users etc

Shard

each shard has the same schema, but data are specific to shard- user data is allocated to specific shard - hash function is used to find the shard e.g. user_id % 4

Sharding key vs partition key

Sharding key vs partition key

Two Problems with sharding

Resharding data - r.g. One shard goes down, Celebrity problem = hotspot key problem

Nanosecond

10^-9

Microsecond

10^-6

Millisecond

10^-3

2^10

1KB ~ 1 000

2^20

1MB ~ 1 000 000

2^30

1GB ~ 1 000 000 000

2^50

1PB ~ 1 000 000 000 000 000

L1 cache

A memory bank built into the CPU chip. Also known as the "primary cache," an L1 cache is the fastest memory in the computer and closest to the processor.

Fan-out service

is a messaging pattern used to model an information exchange that implies the delivery (or spreading) of a message to one or multiple destinations possibly in parallel

Amazon sns

simple notification service

Amazon EC2 - expand

Amazon Elastic Compute Cloud

Amazon EC2 - describe

allows users to rent virtual computers on which to run their own computer applications

Amazon EBS - expand

Amazon Elastic Block Store

Amazon EBS - expand

block level storage volumes for use with EC2 instances. You can create a file system on top of these volumes, or use them in any way you would use a block device (such as a hard drive)

Amazon DynamoDB

is a fully managed proprietary NoSQL database service that supports key–value and document data structures

Amazon Aurora

is a relational database service developed and offered by Amazon Web Services beginning in October 2014 Aurora is available as part of the Amazon Relational Database Service (RDS).

Amazon sqs

simple queue service

memcached

Memcached is an in-memory key-value store for small chunks of arbitrary data (strings, objects) from results of database calls, API calls, or page rendering.

2^40

1 000 000 000 000 = 1TB

L1 cache reference

0.5 ns

L2 cache reference

7 ns

Mutex lock/unlock

100 ns

Main memory reference

100 ns

Compress 1KB with zippy

10 microseconds = 10000 ns

Send 1KB via 1Gbps network

10 microseconds

read 1MB sequentially from memory

250 microseconds

read 1MB sequentially from disk

30 ms

read 1MB sequentially from network

10 ms

sharding vs partitioning

The difference is that sharding implies the data is spread across multiple computers while partitioning does not. Partitioning is about grouping subsets of data within a single database instance

consistent hashing

consistent hashingConsistent hashing is a special kind of hashing such that when a hash table is re-sized and consistent hashing is used, only k/n keys need to be remapped on average, where k is the number of keys, and n is the number of slots

consistent hashing describe algorithm

To determine which server a key is stored on, we go clockwise from the key position on the ring until a server is found.