Using the SDACK Architecture to Build a Big Data Product
3 likes2,008 views
You definitely have heard about the SMACK architecture, which stands for Spark, Mesos, Akka, Cassandra, and Kafka. It’s especially suitable for building a lambda architecture system. But what is SDACK? Apparently it’s very much similar to SMACK except the “D" stands for Docker. While SMACK is an enterprise scale, multi-tanent supported solution, the SDACK architecture is particularly suitable for building a data product. In this talk, I’ll talk about the advantages of the SDACK architecture, and how TrendMicro uses the SDACK architecture to build an anomaly detection data product. The talk will cover: 1) The architecture we designed based on SDACK to support both batch and streaming workload. 2) The data pipeline built based on Akka Stream which is flexible, scalable, and able to do self-healing. 3) The Cassandra data model designed to support time series data writes and reads.
Ad
More Related Content
What's hot (20)
Viewers also liked (20)
Ad
Similar to Using the SDACK Architecture to Build a Big Data Product (20)
Ad
More from Evans Ye (20)
Recently uploaded (20)
![Get & Download Wondershare Filmora Crack Latest [2025]](https://cdn.slidesharecdn.com/ss_thumbnails/revolutionizingresidentialwi-fi-250422112639-60fb726f-250429170801-59e1b240-thumbnail.jpg?width=560&fit=bounds)
Using the SDACK Architecture to Build a Big Data Product
- 1. Yu-hsin Yeh (Evans Ye) Apache Big Data NA 2016 Vancouver Using the SDACK Architecture to Build a Big Data Product
- 2. Outline • A Threat Analytic Big Data product • The SDACK Architecture • Akka Streams and data pipelines • Cassandra data model for time series data
- 3. Who am I • Apache Bigtop PMC member • Software Engineer @ Trend Micro • Develop big data apps & infra
- 4. A Threat Analytic Big Data Product
- 5. Target Scenario InfoSec Team Security information and event management (SIEM)
- 6. Problems • Too many log to investigate • Lack of actionable, prioritized recommendations
- 7. Splunk ArcSight Windows event Web Proxy DNS … AD Anomaly Detection Expert Rule Filtering Prioritization Threat Analytic System Notification
- 8. Deployment
- 9. On-Premises InfoSec Team Security information and event Anomaly Detection Expert Rule Filtering Prioritization
- 10. On-Premises InfoSec Team Anomaly Detection Expert Rule Filtering Prioritization AD Windows Event Log DNS …
- 11. On the cloud? InfoSec Team AD Windows Event Log DNS …
- 12. There’re too many PII data in the log
- 13. System Requirement • Identify anomaly activities based on data • Lookup extra info such as ldap, active directory, hostname, etc • Correlate with other data • EX: C&C connection followed by an anomaly login • Support both streaming and batch analytic workloads • Run streaming detections to identify anomaly in timely manner • Run batch detections with arbitrary start time and end time • Scalable • Stable
- 14. No problem. I have a solution. How exactly are we going to build this product? PM RD
- 15. SMACK Stack Toolbox for wide variety of data processing scenarios
- 16. SMACK Stack • Spark: fast and general purpose engine for large-scale data processing scenarios • Mesos: cluster resource management system • Akka: toolkit and runtime for building highly concurrent, distributed, and resilient message-driven applications • Cassandra: distributed, highly available database designed to handle large amounts of data across datacenters • Kafka: high-throughput, low-latency distributed pub-sub messaging system for real-time data feeds Source: http://www.slideshare.net/akirillov/data-processing-platforms-architectures-with-spark-mesos-akka-cassandra-and-kafka
- 17. Medium-sized Enterprises InfoSec Team AD Windows Event Log DNS …
- 18. No problem. I have a solution. Can’t we simplify it? PM RD
- 19. SDACK Stack
- 20. SDACK Stack • Spark: fast and general purpose engine for large-scale data processing scenarios • Docker: deployment and resource management • Akka: toolkit and runtime for building highly concurrent, distributed, and resilient message-driven applications • Cassandra: distributed, highly available database designed to handle large amounts of data across datacenters • Kafka: high-throughput, low-latency distributed pub-sub messaging system for real-time data feeds Source: http://www.slideshare.net/akirillov/data-processing-platforms-architectures-with-spark-mesos-akka-cassandra-and-kafka
- 21. • Three key techniques in Docker • Use Linux Kernel’s Namespaces to create isolated resources • Use Linux Kernel’s cgroups to constrain resource usage • Union file system that supports git-like image creation Docker - How it works
- 22. • Lightweight • Fast creation • Efficient to ship • Easy to deploy • Portability • runs on *any* Linux environment The nice things about it
- 23. Medium-sized Enterprises InfoSec Team AD Windows Event Log DNS …
- 24. Large Enterprises InfoSec Team AD Windows Event Log DNS …
- 25. Fortune 500 InfoSec Team AD Windows Event Log DNS …
- 27. DB a d d d i w e b Actor Actor Actor … … Log Web Portal Detection ResultAPI Server
- 29. Docker Compose kafka: build: . ports: - “9092:9092” spark: image: spark port: - “8080:8080” …… Web Portal API Server
- 30. Akka Streams and data pipelines
- 31. receiver
- 32. receiver buffer
- 35. • Each streaming job occupies at least 1 core • There’re so many type of log • AD, Windows Event, DNS, Proxy, Web Application, etc • Kafka offset loss when code changed • No back-pressure (back then) • Estimated back-pressure available in Spark 1.5 • Integration with Akka based info lookup services Why not Spark Streaming?
- 36. Akka • High performance concurrency framework for Java and Scala • Actor model for message-driven processing • Asynchronous by design to achieve high throughput • Each message is handled in a single threaded context (no lock, synchronous needed) • Let-it-crash model for fault tolerance and auto-healing • Clustering mechanism available The Road to Akka Cluster, and Beyond
- 37. Akka Streams • Akka Streams is a DSL library for doing streaming computation on Akka • Materializer to transform each step into Actor • Back-pressure enabled by default Source Flow Sink The Reactive Manifesto
- 38. Reactive Kafka • Akka Streams wrapper for Kafka • Commit processed offset back into Kafka • Provide at-least-once message delivery guarantee https://github.com/akka/reactive-kafka
- 39. Message Delivery Sematics • Actor Model: at-most-once • Akka Persistence: at-least-once • Persist log to external storage (like WAL) • Reactive Kafka: at-least-once + back-pressure • Write offset back into Kafka • At-least-once + Idempotent writes = exactly-once
- 40. Implementing Data Pipelines kafka source manual commit Parsing Phase Storing Phase Akka Stream Reactive Kafka
- 41. Scale Up kafka source manual commit Parsing Phase Storing Phase map Async router parser parser parser … scale up
- 42. Scale Out • Scale out by creating more docker containers with same Kafka consumer group
- 43. Be careful using Reactive Kafka Manual Commit
- 44. A bug that periodically reset the committed offset kafka source manual commit #partition offset partition 0 50 partition 1 90 partition 2 70 0 1 2
- 45. A bug that periodically reset committed offset kafka source manual commit kafka source manual commit #partition offset partition 0 300 partition 1 250 partition 2 70 0 1 2 #partition offset partition 2 350
- 47. Abnormal: redline(lag) keep getting reset
- 48. PR merged in 0.8 branch Apply patch if version <= 0.8.7 https://github.com/akka/reactive-kafka/pull/144/commits
- 49. Akka Cluster • Connects Akka runtimes into a cluster • No-SPOF • Gossip Protocol • ClusterSingleton library available
- 50. Lookup Service Integration Akka Streams Akka Streams Akka Streams ClusterSingleton Ldap Lookup Hostname Lookup Akka Cluster
- 51. Lookup Service Integration Akka Streams Akka Streams Akka Streams ClusterSingleton Ldap Lookup Hostname Lookup Akka Cluster
- 52. Lookup Service Integration Akka Streams Akka Streams Akka Streams ClusterSingleton Ldap LookupHostname Lookup Akka Cluster Service Discovery
- 53. Async Lookup Process Akka Streams ClusterSingleton Hostname Lookup Cache Synchronous Asynchronous
- 54. Async Lookup Process Akka Streams ClusterSingleton Hostname Lookup Cache Synchronous Asynchronous
- 55. Akka Streams Cache Akka Streams Cache Async Lookup Process Akka Streams ClusterSingleton Hostname Lookup Cache Synchronous Asynchronous
- 56. Cassandra Data Model for Time Series Data
- 58. Problem • We need to do analytics based on event time instead of processing time • Example: modelling user behaviour • Log may arrive late • network slow, machine busy, service restart, etc • Log arrived out-of-order • multiple sources, different network routing, etc
- 59. Problem • Spark streaming can’t handle event time ts 12:03 ts 11:50 ts 12:04 ts 12:01 ts 12:02
- 60. • Store time series data into Cassandra • Query Cassandra based on event time • Get most accurate result upon query executed • Late tolerance time is flexible • Simple Solution
- 61. Cassandra Data Model • Retrieve a entire row is fast • Retrieve a range of columns is fast • Read continuous blocks on disk (not random seek)
- 62. Data Model for AD log use shardId to distribute the log into different partitions
- 63. Data Model for AD log partition key to support frequent queries for a specific hour
- 64. Data Model for AD log clustering key for range queries
- 65. Wrap up
- 66. • Spark: streaming and batch analytics • Docker: ship, deploy and resource management • Akka: resilient, scalable, flexible data pipelines • Cassandra: batch queries based on event data • Kafka: durable buffer, streaming processing Recap: SDACK Stack
- 67. Questions ? Thank you !