Upload
hazelcast
View
57
Download
1
Tags:
Embed Size (px)
Citation preview
© 2015 Hazelcast Inc.
It’s Time to Make the Move to In-Memory Data GridsFeaturing Greg Luck, Forrester Research’s Mike Gualtieri, and Ellie Mae’s Ken Kolda
© 2015 Hazelcast Inc.
Featuring
2
GREG LUCKCEO, HAZELCAST
MIKE GUALTIERIPRINCIPAL ANALYST, FORRESTER RESEARCH
KEN KOLDASOFTWARE ARCHITECT, ELLIE MAE
Make The Move To In-Memory Data Grids
Mike Gualtieri, Principal Analyst
July 16, 2015
Twitter: @mgualtieri
Planning, implementing, or expanding the use of in-memory data platform.
73%
Base: 1,805 global data and analytics decision-makers Source: Forrester Global Business Technographics Data And Analytics Online Survey, 2015
#Priority
© 2015 Forrester Research, Inc. Reproduction Prohibited 6
52%
53%
53%
54%
58%
64%
64%
65%
66%
73%
75%
0% 10% 20% 30% 40% 50% 60% 70% 80%
Better leverage big data and analytics in business decision-
Create a comprehensive strategy for addressing digital
Create a comprehensive digital marketing strategy
Better comply with regulations and requirements
Improve differentiation in the market
Increase influence and brand reach in the market
Address rising customer expectations
Improve our ability to innovate
Reduce costs
Improve our products /services
Improve the experience of our customers
Customer experience is a top business priority over the next 12 months
Base: 3,005 global data and analytics decision-makers
Source: Global Business Technographics Data And Analytics Online Survey, 2015
For you For all For segments For you
CRM
Hyper-Personal, Real-Time Digital
Experiences
Personal Relationships
Mass Production
Cus
tom
er E
xper
ienc
e
1800 1900 1950 2000 2015
#Celebrity
Customers want and increasingly expect to be treated like celebrities.
• Use analytics to learn customer characteristics and behavior
• Detect real-time context • Adapt applications to serve
an individual customer
Celebrity experiences must:
Be Blazing Fast
#In-Memory
Ubiquitous, near-zero latency for even the most complex data and compute operations.
DEFINITION
FORRESTER Technologies that are principally architected to
use chip-based memory to accelerate the performance of data access and applications;
and reduce the complexity of app development.
Scale should not limit design decisions.
1100
1001
1011
001
0100
1001
1011
001
0100
1100
1101
101
0100
1001
1011
001
His
toric
al
Tran
sact
ions
Cus
tom
er d
ata
Sec
urity
The performance vagaries of accessing data silos can be eliminated.
Fault-tolerance is non-negotiable.
Confidential information must be secure.
In-memory must fit and work seamlessly with existing architectures.
In-Memory technology speeds application development by reducing architectural concerns.
#Priority
73%
#Opportunity
What if you had ubiquitous, near-zero
latency for even the most complex data and
compute operations?
#DataGrids
© 2015 Forrester Research, Inc. Reproduction Prohibited 25
G GG
S S S
Streaming Analytics
B B B
General-purpose data processing cluster
D
Scale-up Database
Data And Compute Grid
D D D
Clustered Database
#UseCases
Overcome legacy architectural bottlenecks such as databases.
Real-time integration for shared cache.
Fast distributed processing.
HAZELCAST @ ELLIE MAE
Ken Kolda SOFTWARE ARCHITECT
About Ellie Mae
• Founded in 1997 to provide soDware soluFons to mortgage industry
• Released Encompass loan originaFon system in 2003 • Boxed, on-‐premise soDware sold in retail stores • Target market was 2 – 20 user mortgage broker
• Started offering Hosted model in 2007 • Relieved customers from burden of IT infrastructure, processes • Roughly 80% of customers currently use Hosted model, with customers up
to 3000 users
• Began engineering of “Next Gen” soluFon in 2013 • Hybrid cloud model for truly SaaS soluFon
7/22/15 ©2015 Ellie Mae. All rights reserved. 32
7/22/15 ©2015 Ellie Mae. All rights reserved. 33
The Problem of Scale
• Legacy Architecture • .NET-‐based client/server applicaFon with
SQL Server back-‐end • Designed for on-‐premise, single-‐tenant
deployment
• Scalability Limits • Server adapted to support limited mulF-‐
tenancy • Home-‐grown, in-‐process caching added
to ease load on SQL Server • Cache synchronizaFon issues prevent
horizontal scale, load balancing
DataStore
Encompass Clients
Write Read
Encompass Server
Cache
Encompass Server
Cache
Encompass Server
Cache
DataStore
Encompass Clients
Write Read
Encompass Server
Encompass Server
Encompass Clients
Write Read
DataStore
7/22/15 ©2015 Ellie Mae. All rights reserved. 34
Horizontal Scale via Hazelcast
• Distributed Cache Tier • In-‐process cache replaced with
shared Hazelcast data grid • Hazelcast locking used to ensure
transacFonal consistency between server nodes
• Scalability & Availability • App Fer can now scale horizontally • EliminaFon of single point of failure • App servers can be brought up/
down without loss of cache • Hazelcast cluster can be expanded
to meet future demands
Encompass Server
Encompass Server
Encompass Clients
Write Read
DataStore
7/22/15 ©2015 Ellie Mae. All rights reserved. 35
Choosing Hazelcast
• Requirements/Use cases • Database caching: Cross-‐process locks, Hibernate second-‐level caching • Session storage: TTL and Idle Time support, evicFon/expiraFon noFficaFons
& policies • NaFve support for .NET, Java clients; REST support • High-‐performance, highly available, horizontally scalable • Enterprise support
• EvaluaFon process • High-‐level evaluaFon of 8 in-‐memory data grids (JDG, Couchbase, JvCache,
Terracona, Riak, Redis, Memcached) • Six-‐month PoC on top three contenders (Hazelcast, JDG, JvCache) • Rated on criteria including performance, fault tolerance, client support,
ease-‐of-‐configuraFon/operaFon, monitoring, noFficaFons
7/22/15 ©2015 Ellie Mae. All rights reserved. 36
EvaluaFon Results
* Note: Evalua-on performed using Hazelcast 3.0
Thank You!