The document discusses Aerospike, an enterprise-class NoSQL database designed for real-time applications, emphasizing its hybrid memory architecture that provides enhanced speed, scalability, and lower total cost of ownership (TCO) compared to traditional databases. It highlights numerous business cases demonstrating how Aerospike effectively addresses high transaction loads, low latency, and the need for reliability while supporting complex data requirements. The presentation showcases the advantages of Aerospike in various sectors including finance, telecommunications, and ad tech.

1. 1VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Aerospike: The Enterprise Class
NoSQL Database for Real-Time
Applications
Srini V. Srinivasan
Founder, Chief Development Officer
Brillix-Aerospike Event
Tel Aviv
June 14, 2017

2. 2VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Real-time Workloads

3. 3VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Billion Dollar Advertising Market uses RTB
1 to 6 billion cookies tracked
Auctions at about 3.0M / sec in North America
100ms ad rendering, 50ms real-time bidding
Low Latency, High Throughput, High Uptime

4. 4VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
LEGACY DATABASE
(Mainframe)
XDR
Decisioning Engine
DATA WAREHOUSE/
DATA LAKE
LEGACY RDBMS
HDFS BASED
BUSINESS
TRANSACTIONS
Web views
( Payments )
( Mobile Queries )
( Recommendation )
( And More )
High Performance NoSQL
“REAL-TIME BIG DATA”
“DECISIONING”
500
Business Trans per sec
5000
Calculations per sec
X = 2.5 M
Database Transactions per sec
Operational Scale in Enterprises

5. 5VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Hybrid Memory
Architecture

6. 6VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Traditional architecture has significant limitations
Challenges
• Complex
• Maintainability
• Durability
• Consistency
• Scalability
• Cost ($)
• Data Lag
Caching Layer
Operational Database
Real-time
Consumer Facing
Pricing /
Inventory/Billing
Real-time
Decisioning
Streaming
Data
Legacy Database
(Mainframe)
RDBMS
Database
Transactional
Systems
Enterprise Environment
Legacy RDBMS
HDFS BASED
Fast speed – Consumer Scale

7. 7VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
NoSQL DB
Hybrid Memory Architecture
Benefits
• Fast App
Development
• Richer data schema
• No need for SQL
• In-memory
performance
• High scale
• Lower latency
• Distributed
• Tradeoff:
Consistency versus
Availability
Caching Layer
Real-time
Consumer Facing
Pricing /
Inventory/Billing
Real-time
Decisioning
Streaming
Data
Legacy Database
(Mainframe)
RDBMS
Database
Transactional
Systems
Enterprise Environment
Legacy RDBMS
HDFS BASED
Fast speed – Consumer Scale

8. 8VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Architecture – The Big Picture
1) No Hotspots
– Distributed Hash Table
simplifies data partitioning
2) Smart Client – 1 hop to data,
no load balancers
5) Transactionsand long-
runningtasks prioritized in
real-time
6) XDR – sync replication across
data centers ensures Zero
Downtime
3) Shared NothingArchitecture,
every nodeis identical
4) Smart Cluster, Zero Touch
– auto-failover, rebalancing,
rack aware, rolling upgrades

9. 11VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Aerospike Storage Architecture
1. Direct device access
2. Large Block Writes
3. Indexes in DRAM
4. Highly Parallelized
5. Log-structured FS “copy-on-write”
6. Fast restart with shared memory
Optimized for SSDs
Storage Layout
Hybrid Memory In-Memory

10. 12VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Hybrid Memory Benefits

11. 13VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Bare metal performance
YCSB Benchmark
50 million records
YCSB Workload A (50/50 R/W)
YCSB Workload B (95/5 R/W)
Zipfian key distribution
8 Core Dual Socket Intel®
Xeon ® CPU E5-2665@2.4GHz
32GB DRAM with 16 queues

12. 14VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
DRAM Vs SSD on GCE
GCE instance n1-standard-8
10 node cluster
150 byte record with 3 columns
100 million records

13. 15VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
1 32 4 5
Phases
1) 100KTPS – 4 nodes
2) Clients at Max
3) 400KTPS – 4 nodes
4) 400KTPS – 3 nodes
5) 400KTPS – 4 nodes
Aerospike Node Specs:
CentOS 6.3
Intel i5-2400@ 3.1 GHz (Quad core)
16 GB RAM@1333 MHz
Predictable Performance During Failures

14. 16VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
In Memory vs. Hybrid Memory – SLA in Actual
Deployment
In-Memory SLA 98.5% Hybrid Memory SLA 99.95%
Missed SLA is lost Revenue!!!

15. 17VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
In-Memory vs Hybrid Memory – Low TCO @ Scale
Year 1
(Millions USD)
Year 2
(Millions USD)
Year 3
(Millions
USD)
Total
(Millions USD)
Cassandra $1.82 $2.53 $3.94 $8.28
Aerospike $1.88 $1.24 $1.85 $4.97
Total OpEx
Savings
-$0.06 $1.28 $2.09 $3.32

16. 18VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Hybrid Memory Advantage – Lower TCO & better SLA
Aerospike’s hybrid memory has an ‘Unfair Competitive Advantage’ vis a vis traditional “In Memory” Solutions
Customer Situation Problem
Aerospike Hybrid
Memory
Trading Account Risk
Management
DB2+Gemfire cache 150 Servers growing to 1000
Single Aerospike cluster – 12
servers
Payments Fraud Detection
2 ORCL RAC clusters +
Terracotta cache
System Stability & missing
SLA’s
3 Aerospike Clusters – 20
Servers each
User Integrity Checking
for Internet Transactions
DataStax/Cassandra
168 DataStax Servers growing
to 450+
30 Aerospike Servers –
2 clusters
Telco Device and User
Access
ORCL Coherence / DataStax
Cassandra
Existing SOE solutions unstable
& Costly
5 successful POC’s against
Cassandra & ORCL Coherence
Telco Revenue Assurance
DataStax/Cassandra
PostgreSQL + cache
Hundreds of cache &
Cassandra Servers
Scalability challenges
Significant reduction of server
footprint – global deployment

17. 19VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Next Generation Systems of Engagement –
An Emerging Market with Multiple Technologies
Hybrid Memory Architecture Delivers Predictable Performance, Highest Availability, and Lowest TCO
Systems of Engagement - TCO
TCO
($)
Scale TB
Systems of Engagement – Many Choices
Alternative
TCO
Aerospike
TCO
Speed
TPS
Scale TB
Significant functional
overlap - Commodity DB
problem set
Unique Functional
Capabilities and High
Value Problem Set

18. 20VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Thank You
Questions?

19. 21VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Appendix: Case Studies

20. 22VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Business Challenge
■ Must update stock prices, show balances on 300 positions, process
250M transactions, 2 M updates/day
■ High access from mobile was killing the DB2 Database under
normal transaction load
■ Calculate risk metrics on portfolios on a continuous basis
Caching solution failed
■ Running out of memory, data inconsistencies, restarts take 1 hour
■ 150 Servers -> Growing to 1000
Need to scale business
■ 3 → 13 TB, 100 → 400 Million objects, 200k → I Million TPS
Aerospike Hybrid Memory Advantage
■ Built for Flash – eliminated inconsistencies
■ Predictable Low latency at High Throughput – handled mobile
access easily for enhanced transaction load
■ Cross data center (XDR) support provides high availability
■ 10-12 Server Cluster – reduced from 150 in-memory cache servers IBM DB2
(MAINFRAME)
Read/Write
Start of Day
Data Loading
End of Day
Reconciliation
Query
REAL-TIME
DATA
FEED
ACCOUNT
POSITIONS
XDR
Retail Banking Positions – Risk Management

21. 23VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
CREDIT CARD
PROCESSING SYSTEM
FRAUD DETECTION &
PROTECTION APP
ACCOUNT
BEHAVIOR
ACCOUNT
STATISTICS
STATIC DATA
RULE 1 – PASSED ✔
RULE 2 – PASSED ✔
RULE 3 – FAILED ✗
HISTORICAL
DATA
RULES
RULE 1
RULE 2
RULE 3
…
BusinessChallenge
■ Every payment transaction requires
hundreds of DB reads/writes
■ Missed latency SLA lost business
■ Caching solution too expensive
Need to scale up
■ 10 → 100 TB
■ 10B → 100 B objects
■ 200k → I Million+ TPS
Selected Aerospike
■ Built for Flash – eliminated inconsistencies
■ Predictable Low latency at High Throughput
■ Cross data center (XDR) support provides high availability
■ 20 Server Cluster reduced from 150 in-memory cache servers
■ Used latest technology to reduce cost – Dell 730xd w/ 4NVMe SSDs
Fraud Prevention for Interactive Payments

22. 24VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
BusinessChallenge
• Edge access to regulate traffic on network
• Accessible using provisioning applications
(self-serve and through support personnel)
• Ensure accuracy in billing and charging
• Quick turn around for provisioning changes
• Existing solutions were failing at scale
Need Extremely High Availability,
Reliability, Low latency
• > TBs of data
• 10-100M objects
• 10-200K TPS
Selected Aerospike
• Clustered system – easy to deploy and maintain
• Predictable low latency at high throughput
• Highly-available and reliable on failure
• Cross data center (XDR) support
SOURCE
DEVICE/USER DESTINATION
Real-Time
Auth. QoS Billing
Request Execute
Request
Real-Time ChecksConfig Module App
Update Device
User Setting
Hot-Standby
XDR
Telco – Revenue Assurance – Billing and Charging

23. 25VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Ad tech - HTAP – Click Fraud Avoidance
DATA WAREHOUSE/
DATA LAKE
BATCH BI
Actionable insight
& Refresh
40 TB Data
SLA: A few
minutes
Real-time Ad Serving
40TB data per cluster
Deep Analysis
Historical data
XDR
XDR
XDR
XDR

24. 26VLDB 2016, Delhi, India. © 2016 Aerospike Inc. All rightsreserved.[ ]
Fraud Detection and Revenue Assurance – Systems of
Engagement
Algorithmic
Decisioning Engine
In-Memory/Hybrid-Memory
Database
DATA WAREHOUSE/
DATA LAKE
Actionable
Insight
Transactional Analytics
BATCH BI
MACHINE
LEARNING
REAL-TIME BI
Fraud Prevention
Call Detail Record
Gaming
Banking
Payment Gateway
Real-time Bidding
Packet Routing
Msg / Chat
Look & Book
eCommerce
Session Management
IoT
Low Read Latency
Heavy Write Load
100% Up Time