Deploying Microservices with Kubernetes and Redis

1. Microservices: Is it time to
break up?
DAVE NIELSEN | @DAVENIELSEN | DAVE@REDISLABS.COM
HEAD OF ECOSYSTEM PROGRAMS
REDIS LABS

2. In-memory Multi-model Database
2

3. Optionally Persistent
3

4. Redis Top Differentiators
Simplicity ExtensibilityPerformance
NoSQL Benchmark
1
Redis Data Structures
2 3
Redis Modules
4
Lists
Hashes
Bitmaps
Strings
Bit field
Streams
Hyperloglog
Sorted Sets
Sets
Geospatial Indexes

5. Performance: The Most Powerful Database
Highest Throughput at Lowest Latency
in High Volume of Writes Scenario
Least Servers Needed to
Deliver 1 Million Writes/Sec
Benchmarks performed by Avalon Consulting Group Benchmarks published in the Google blog
5
1
Serversusedtoachieve1Mwrites/sec

6. Redis – First Cloud Native Database
1. Client writes data to Master in-memory
2. Master writes data to Replica in-memory for High Availability
• Then Master responds to the Client
• If Master fails, then Replica takes over
3. Replica saves data to disk for Disaster Recovery
• (attached storage or object store)
• If disaster occurs, Master & Replica recover from disk
6
2. Replica1. Master
Client
3. Changes

7. “REDIS IS FULL OF DATA STRUCTURES!”
2 Simplicity: Data Structures - Redis’ Building Blocks

8. Simplicity: Redis Data Structures – ’Lego’ Building Blocks
Lists
[ A → B → C → D → E ]
Hashes
{ A: “foo”, B: “bar”, C: “baz” }
Bitmaps
0011010101100111001010
Strings
"I'm a Plain Text String!”
Bit field
{23334}{112345569}{766538}
Key
8
2
”Retrieve the e-mail address of the user with the highest
bid in an auction that started on July 24th at 11:00pm PST” ZREVRANGE 07242015_2300 0 0=
Streams
{id1=time1.seq1(A:“xyz”, B:“cdf”),
d2=time2.seq2(D:“abc”, )}
Hyperloglog
00110101 11001110
Sorted Sets
{ A: 0.1, B: 0.3, C: 100 }
Sets
{ A , B , C , D , E }
Geospatial Indexes
{ A: (51.5, 0.12), B: (32.1, 34.7) }

9. 9
University Style
• 4-6 week course duration
• Tutorials, Online Labs, Homework
• 3 hours / week commitment (20 hours
per course)
• Certificate of Completion
university.redislabs.com
Free, online education
for the Redis Comminity
Next Courses start Nov-27
• RU101 Data Structures
• RU201 RediSearch
In Development
• RU102J Redis for Java Devs
• RU202 Redis Streams

10. Monolith or Microservices?

11. Benefits of Microservices
• Make it perform faster or scale better
• Extend an application’s capabilities more easily
• Add new features more quickly and easily
• Improve maintainability
• Reduce vulnerabilities
11

12. However, Microservices are Complicated
12
• More than meets the eye

13. Be Prepared for Success Now, and Later
• What to do when app begins to hockey stick
• Duck tape the parts when they break?
• Do you rewrite your app with scalability in mind?
13

14. Do Both with Data Services + Microservices
• Data Services - Redis solves
web scale data problems
14
• Microservices - Kubernetes
solves the hockey stick problem

15. Scale with Redis and Microservices
• In many cases, Monolith plus Data Services is best
• Smaller apps and small teams don’t need overhead and
unnecessary complexity of Microservices Architecture
• Then, when its time to scale, use Microservices with Redis
15

16. Scale with Redis Data Services while Re-
architecting for Microservices Architecture

17. Scale with Redis Data Services
… while Re-architecting for Microservices Architecture
17
Top Redis Data Services
• Cache Long Running Queries
• User Session Management
• Queues/Data Ingest
• Streaming
• Custom Analytics / Leaderboard
• Pub-Sub

18. User Session Store
• The Problem
• Maintain session state across
multiple servers
• Multiple session variables
• High speed/low latency required
Why Redis Rocks
• Hashes are perfect for this!
• HSET lets you save session
variables as key/value pairs
• HGET to retrieve values
• HINCRBY to increment any
field within the hash structure

19. Redis Hashes Example
userid 8754
name dave
ip 10:20:104:31
hits 1
lastpage home
hash key: usersession:1
HMSET usersession:1 userid 8754 name dave ip 10:20:104:31 hits 1
HMGET usersession:1 userid name ip hits
HINCRBY usersession:1 hits 1
HSET usersession:1 lastpage “home”
HGET usersession:1 lastpage
HDEL usersession:1 lastpage
Hashes store a mapping of keys to values – like a dictionary or associative array – but faster
DEL usersession:1

20. Managing Queues of Work
• The Problem
• Tasks need to be worked on asynch
to reduce block/wait times
• Lots of items to be worked on
• Assign items to worker process and
remove from queue at the same time
• Similar to buffering high speed data-
ingestion
Why Redis Rocks
• Lists are perfect for this!
• LPUSH, RPUSH add values at
beginning or end of queue
• RPOPLPUSH – pops an item
from one queue and pushes it
to another queue

21. Redis Lists Example
LPUSH queue1 orange
LPUSH queue1 green
LPUSH queue 1 blue
RPUSH queue1 red
LPUSH adds values to head of list
RPUSH adds value to tail of list
blue green orange .. .. red

22. Redis Lists Example
blue green orange .. ..
RPOPLPUSH queue1 queue2
red
LPUSH queue1 orange
LPUSH queue1 green
LPUSH queue1 blue
RPUSH queue1 red
RPOPLPUSH pops a value from one list and pushes it to another list

23. Leaderboard with Sorted Sets Example
• The Problem
• MANY users playing a game or
collecting points
• Display real-time leaderboard.
• Who is your nearest competition
• Disk-based DB is too slow
Why Redis Rocks
• Sorted Sets are perfect!
• Automatically keeps list of
users sorted by score
• ZADD to add/update
• ZRANGE, ZREVRANGE to get
user
• ZRANK will get any users
rank instantaneously

24. Redis Sorted Sets
ZADD game:1 10000 id:1
ZADD game:1 21000 id:2
ZADD game:1 34000 id:3
ZADD game:1 35000 id:4
ZADD game:1 44000 id:3
or
ZINCRBY game:1 10000 id:3
34000 id:3
35000 id:4
21000 id:2
10000 id:1
ZREVRANGE game:1 0 0
ZREVRANGE game:1 0 1 WITHSCORES
44000 id:3
+ 10000id:3

25. Examples of Data Services
• Traditional Cache
• Smartcache or write through cache
• User Sessions Management
• Leaderboard (Netflix API use case)
• Who’s Online (Pinterest use case)
• Tag Server (Stack Overflow use case)
25

26. Scale with Microservices Architecture

27. Microservices at Netflix

28. Break App Up into a Microservice
• Large Companies: Teams gets too big. Map team to org structure.
• Conways Law: “Organizations which design systems… are
constrained to produce designs which are copies of those
organizations”
• Reverse Conways Law: “Create your org structure to map the
Microservices in your application”
• Startups: Break apart for technical reasons
• Parts of your app needs to scale at different rates
• You want to use different programming languages
28

29. Examples of Microservices with Redis: By Department
• Product Catalog
• Product Recommendations
• User Checkout and Order Processing
• Order Fulfillment
• Customer Loyalty
29

30. Examples of Microservices with Redis: By Technology
• Email Campaigns - Sending 25k Emails (ex: SendGrid)
• Other User Communications: (ex: email, text, web sockets, etc.)
• URL Shortener (ex: Bit.ly)
• Webcache: Scrape and Cache (ex: Twitter)
• Machine Learning/Deep Learning (Scikit-Learn/Tensorflow)
• “Autocomplete” or Search-as-you-Type (RediSearch)
30

31. • https://github.com/dnielsen/uploadvalidate/blob/master/website-pod.yaml
• spec:
–containers:
• env:
–name: API_IP
–value: api # Service Discovery using pod label
Service Discovery
31
The simple Kubernetes way

32. Redis can do Many Things
32

33. Real Time
Analytics
User Session
Store
Real Time Data
Ingest
High Speed
Transactions
Job & Queue
Management
Time Series Data Complex
Statistical Analysis
Notifications Distributed Lock Content Caching
Geospatial Data Streaming Data Machine Learning
Redis can do many things
33
Very Large Data Sets Search

34. High Availability
• Cover your business for peak hour traffic
• Prevent cache stampede
• Deliver consistent high performance
• Zero downtime scaling
Security
• Meet your organization’s security and
compliance requirements
Performance
• Maximize resource utilization: Run on all CPU
cores
Advantages of Redis Enterprise
34
Active-Active
• Maintain consistent session state across all data
centers
• CRDT based active-active delivers strong
eventual consistency with local latency
Redis on Flash
• No need to evict or delete session data; allow
the data to overflow to Flash memory

35. For any enterprise
datacenter or cloud
environment
Fully managed, server-
lessly scaling service in
the VPCs of public
cloud platforms
Fully managed, server-
less Redis Enterprise
service on hosted
resources in the public
cloud platforms
How to Get Started with Redis Enterprise?
35
or or
Cloud VPC Software

36. Thank you!
redislabs.com
36