Maroš Kopec, Bibiána Ťureková, Nikoleta Češeková, Iuliia Gurianova
1
Agenda
● Evolution
● Main features
● Core principles
● Architecture
● Comparison with MongoDB
● Demo
2
3
- Memcached
technology
- Sharding
- Replication
- Persistence
- Document
oriented
model based
on JSON
Main Features
- Flexible data model
- Json documents with no fixed schema
- Easy scalability
- Grow cluster without downtime or application changes
- Automatic sharding and rebalancing
- High performance
- consistently high throughput
- sub-millisecond response times
- Always on
- No downtime for upgrades, maintenance...
4
Developer integration
- Provide libraries for:
- Net
- PHP
- Ruby
- Python
- C
- Node. js
- Java
- GO
5
Key customers
6
Real use cases
● Real-time user activity
● User preferences
● Caching
● Promotions tracking
● Users targeting
7
Core principles
- Documents stored in
buckets (“db”)
- vbuckets
- Owner of keys
- Responsible for
distribution and
replication
- Document = key + value
8
Architecture
- Highly distributed
architecture
- Horizontal scaling by hash
sharding (by document ID)
- Servers are equal
- Server
- Data manager
- Cluster manager
9
Replication
10
- documents in a bucket
are isolated
- configurable number of
replicas (up to three)
- if a server crashes, the
system will locate the
replicas of the
documents, and
promote them to active
status
11
● masterless, clustered, and
replicated distributed database;
Multi-Dimensional Scaling
(MDS)
● caching layer for both data and
indexes
● SQL-based query language
● Web UI, CLI, REST API
● master-slave; Single-node
and sharded environments
● third-party caches
● unique proprietary language
● CLI tools for maintenance
CouchbaseMongoDB
Multi-Dimensional Scaling
12
13
Queries
Update workload
14https://www.datanami.com/2019/01/16/benchmarking-nosql-databases/
Short-range scan workload
15
16
17
DEMO
18
Questions
19