Riak at Engine Yard - given at Big Nerd Ranch

Riak is an open source, highly scalable and fault-tolerant distributed database. Its high availability features in particular make Riak an ideal fit for cloud environments where your resources may disappear without notice but your database must continue to serve requests.

This talk will introduce you to Riak’s use cases and overall architecture. We’ll also cover how we run Riak at Engine Yard.

Transcript

1. Riak in the Cloud Getting Started with Big Nerd Ranch
2. Ines Sombra @RandomMood About Me
3. About Us Our Agenda The Problem The Solution Our Customers
4. Engine Yard
5. About US 100% Cloud Based Critical services MUST be up 24/7/365 Clouds Regions Stacks Support HA DR
6. http://www.flickr.com/photos/meetrajesh/6972883401 Our Customers
7. An APPLICATION App App App DB Master DB Replica us-east-1a us-east-1b
8. http://www.flickr.com/photos/brandongrasley/8227882239/ Then AWS Crashed
9. Our App AFTER App App App DB Master DB Replica us-east-1a us-east-1b X
10. ! This sucked, A LOT Time to strengthen our HA/DR Story Customers and us need to be up all the time! $
11. Relational Document Key/Value Column Data Models Consistency Partitions Our Research Are ACID properties really needed? Consistency Availability
12. Relational Model Impedance Mismatch
13. Memory data-structure Impedance Mismatch
14. We needed a fault tolerant and highly available (clustered) datastore
15. Linearly scalable Fault-tolerant Finally provide HA/DR solution OSS, Dynamo-based, distributed K/V store Hello RIAK!
16. Riak’s Use Cases Application Type Key Value Session User/Session id Data Advertising Campaign id Data Logs Date Log file Content title, integer text, json, xml, media..
17. When is Riak a good fit When availability is more important than consistency When data can be modeled as keys/values When the problem fits
18. Riak is OPS FRIENDLY node 1.2.1 node 1.2.1 node 1.2.1 node 1.2.1 node 1.2.1 Growth Reduction Upgrades Failures node 1.3.0 X
19. World’s most famous slide
20. Riak is SO MUCH MORE Consistent Hashing Hinted Handoff Active Anti Entropy Tuneable consistency N | W | R
21. OUR APP AFTER App App node 1.3.0 node 1.3.0 node 1.3.0 node 1.3.0 node 1.3.0 App
22. We need a Riak Cluster but we have no hardware
23. RIAK In the Wild 18th and Harrison, San Francisco, CA
24. Let’s build A PRODUCT http://www.flickr.com/photos/worldbank/8261699153/
25. 1 2 3 4 Number of Nodes Your CHOICES You are trying to create a Riak cluster. Do you need any help? Instance Type Data Location Backend Type
26. What we GIVE YOU
27. Our SUPPORT X2
28. What you OPS to know HOWTO
29. Prepare Yourself More instances to compensate for the performance variability of virtualized resources
30. ^ m1.large & m1.xlarge EBS-optimized or PIOPs High I/O Quadruple XL (hi1.4xlarge) - SSDs! Choosing INSTANCES
31. 5 nodes or more Best performance & growth Scales linearly Don’t be stingy Cluster Sizes
32. Change Default CANNOT BE RESET (8-64 vnodes per node) Ring SIZE Ring Size at 256 DONT GO OVER 512
33. Our PROVISIONING App haproxy MyRiakCluster: node-0 MyRiakCluster: node-1 MyRiakCluster: node-2 MyRiakCluster: node-3 MyRiakCluster: node-4 Riak Cluster HTTP or Protobuff App haproxy App haproxy Riak Cluster
34. ^ Riak process pages should not hit swap Swap usage can result in unresponsive server Let the kernel kill it Disable SWAP
35. ^ Mount with noatime flag Use deadline scheduler on EBS Mount & SCHEDULER cat /sys/block/xvdj1/queue/scheduler noop [deadline] cfq echo deadline > /sys/block/xvdj1/queue/ scheduler
36. ^ xfs, ext3 ext4 barrier = 0 data = writeback ZFS not recommended Linux FILESYSTEMS
37. ^ Backup both the ring and data directories scp or rsync data dir EBS snapshots Backup STRATEGY
38. Monitoring STRATEGY Stats Interface riak-admin status Collectd Nagios
39. ^ Vertical Scaling RIAK riak-admin cluster replace <old> <new> ^ Horizontal riak-admin cluster add <new>
40. How we SCALE RIAK
41. + PostgreSQL RIAK Hybrid Solutions FTW + FTS Engine* RIAK + Redis RIAK
42. Try OUR RIAK
43. How difficult IS IT?
44. A few thoughts Finally!
45. Pick the right tool for the job Apps & Services can have multiple databases Polyglot PERSISTENCE
46. On SOA My super nice e-commerce site Shopping Cart & Session Data Completed Orders Inventory and Item Pricing Recommendations Engine Session Storage service K/V Store Order Persistence service Document Store Inventory & Price service RDBMS Nodes and relations service Graph Store NoSQL Distilled: Fowler & Sadalage
47. Don’t optimize too quickly Take time to understand your problem Complexity increases with more databases We RECOMMEND
48. Do KEEP IN MIND Relational Document Store != Different paradigms are different
49. Give me MOAR! Whatever
50. Little Riak Book (https://github.com/coderoshi/little_riak_book/) docs.basho.com blog.engineyard.com nosql.mypopescu.com Great RESOURCES
51. Questions?