Distributed Databases and more specifically cloud-native databases were created to face many of the issues with a traditional relational database. Having a low latency and highly available database is the key to preventing a multitude of issues. This talk will focus on what distributed databases provide and why it’s important. This talk will also focus on how cloud-native databases like Apache Cassandra are the perfect match for multi-cloud architectures, and why multi-cloud is important.

1. ©2019 DataStax.
Use only with permission.
1
academy.datastax.com
©2019 DataStax.
Use only with permission.
Cassandra and the Multi-Cloud
Amanda K Moran
Developer Advocate for DataStax

2. But first… A Little About Amanda
• Graduated with MS in Computer Science and
Engineering from Santa Clara University in
2012
• Worked as a Software Engineer for 6 years and
now is a Developer Advocate
• Apache Committer, PMC Member, and initial
contributor to all installation and deployment
work for Apache Trafodion
• Keywords: Disney, Cloud, Dogs, Veggies,
Linux, Databases, Big Data, Analytics, Testing,
and Running
© DataStax, All Rights Reserved.2 Confidential

3. What Are We Talking About Today
• Introduction to Apache Cassandra
• What are Multiple DataCenters?
• Why all this talk about MultiCloud?!
• Apache Cassandra and the MultiCloud
• Demo!
3 © DataStax, All Rights Reserved. Confidential

4. Introduction to Apache
Cassandra
4 © DataStax, All Rights Reserved. Confidential

5. What is Apache Cassandra?
• First developed by Facebook
• Became a top-level Apache
Foundation project in 2010
• NoSQL database
• Distributed, decentralized database
• Elastic scalability -- add/remove nodes with no
downtime
© DataStax, All Rights Reserved. Confidential5

6. What is Apache Cassandra?
• High performance
– Very fast -- low latency
• High availability / fault tolerant
– No single point of failure
• Solves many of the problems faced with a
traditional DB for certain workloads
© DataStax, All Rights Reserved. Confidential6

7. What Does All This Mean?
• Let’s talk about the Big Topics:
– Distributed Systems
– Replication
– Elastically Scalable
– High Availability
– Latency
• Read path
• Write path
© DataStax, All Rights Reserved. Confidential7
Note: Don’t forget this is just a brief intro!

8. Distributed System
• Every node in the cluster has the same role
– Really!
– Cassandra does not have a Master-Worker Architecture
• Any client can connect to any node
– All nodes are Read and Write ready
• But this is not to say that all nodes
contain all data
© DataStax, All Rights Reserved. Confidential8

9. The cluster
9
Server
Token Range
0 0-25
26 26-50
51 51-75
76 76-100
Server
ServerServer
0-25
76-100
26-5051-75

10. Replication
• To be able to survive a node going down data must be
copied to other nodes
• The Replication Factor (RF) is set by the user
– 1-Number of nodes in the Cluster (not recommended)
• The data is asynchronously replicated
– Automatic
– Peer-to-peer communication
© DataStax, All Rights Reserved. Confidential10

11. Replication
11
DC1
DC1: RF=3
Node Primary Replica Replica
10.0.0.1 00-25 76-100 51-75
10.0.0.2 26-50 00-25 76-100
10.0.0.3 51-75 26-50 00-25
10.0.0.4 76-100 51-75 26-50
10.0.0.1
00-25
10.0.0.4
76-100
10.0.0.2
26-50
10.0.0.3
51-75
76-100
51-75
00-25
76-100
26-50
00-25
51-75
26-50

12. Elastically Scalable
• As more nodes are added, performance increases linearly
• You can scale up or down with no downtime
– Not even a restart!
• Reads and Writes both scale
© DataStax, All Rights Reserved. Confidential12

13. High Availability
• The lack of a Master node allows for high availability
– No single point of failure
• Replication allows nodes to fail and data to still be
available
– Cassandra expects nodes to fail and doesn’t panic
© DataStax, All Rights Reserved. Confidential13

14. Latency
• How is Apache Cassandra able to achieve such low
latency?
• It’s all about the read and write path!
– The write path is truly beautiful in its simplicity!
– High throughput with quick responses times are easy to
achieve
© DataStax, All Rights Reserved. Confidential14

15. Write Path (Client to Cluster)
© DataStax, All Rights Reserved. Confidential15

16. Write Path (Internals)
© DataStax, All Rights Reserved. Confidential16

17. Read Path (Client to Cluster)
• Data modeling comes in to play here!
– This is the one “simple trick about Cassandra/Nosql”
• Partition data by nodes
• Query will essentially query one node and return the data
– Constant time READ access
© DataStax, All Rights Reserved. Confidential17
select * from myTable where state =`CA`
Looks like SQL but it IS NOT! It’s
Cassandra Query Language

18. Multiple Data Centers
• Cassandra Cluster represented as a ring
• Can support multiple rings chained together
– Separated by region
– Separated by workload
© DataStax, All Rights Reserved. Confidential18

19. Multiple Data Centers
• Multiple Data Center support is out of the box
• Replication happens between data centers automatically
– No need to sync data
© DataStax, All Rights Reserved. Confidential19

20. What is DataStax?
• DataStax is the enterprise version of Apache Cassandra
• 70% of the the commits to the open source project
• 2x the Write performance of Apache Cassandra
• 2x the Read performance
• Add in the ability to do Search, Analytics, and Grpah
• Cool tools!
© DataStax, All Rights Reserved. Confidential20

21. WHY Multi-Cloud?
21 © DataStax, All Rights Reserved.

22. What is Multi-Cloud?
• Two or more public cloud providers at the same time
– Data moving between two++ providers
© DataStax, All Rights Reserved. Confidential22

23. Why Multi-Cloud?
• Data Center Locality
– Not all zones are in each provider
• Provider Specific Services
• Cloud provider competition
– Can shop around cheap compute! -- Maybe
– More likely -- afraid of lock in to a competitor
• Cost
© DataStax, All Rights Reserved. Confidential23

24. Why can Apache Cassandra do MultiCloud?
• Multi Data Center support -- out of the box
• Cloud Native database
– Built for the cloud
• Multi region support
– Expanded to Hybrid cloud
– Easy expansion to Multi-Cloud
• Every node has the same job
© DataStax, All Rights Reserved. Confidential24

25. Only Database that supports Multi-Cloud
© DataStax, All Rights Reserved. Confidential25

26. Only Database that supports Multi-Cloud
© DataStax, All Rights Reserved. Confidential26

27. Demo
27 © DataStax, All Rights Reserved.

28. Why NOT Multi Cloud?
28 © DataStax, All Rights Reserved.

29. Issues with Multi-Cloud
• Complexity
• Networking!
– Latency
• Security
– Boundary protection
• Legal
• Scaling at the Application layer
© DataStax, All Rights Reserved. Confidential29

30. Okay, this was
awesome! What now?
30 © DataStax, All Rights Reserved.

31. Information and Links
• Learn more about Cassandra: https://academy.datastax.com/
• Learn more about DataStax: https://www.datastax.com/
• Follow me on Twitter: @AmandaK_Data
• Github: https://github.com/amandamoran
© DataStax, All Rights Reserved. Confidential31

32. Join us at Accelerate!!
www.datastax.com/accelerate Discount Code: ADVOCATE20

33. Thank you
33 © DataStax, All Rights Reserved. Confidential