The document discusses SAS Institute, a leading analytics software company. It provides an overview of SAS Institute's products and history, including that it began with IBM Assembler and PL/1 and has since transitioned to languages like C and Java. It also discusses SAS Institute's efforts to make its software compatible with multiple operating systems and cloud platforms. The document explores SAS Institute's investigations into deploying its Visual Analytics and SAS Studio products on Cloud Foundry and the potential benefits and challenges of doing so.
59. 60
Technologies
540+ Products and 5000+ applications, components, services, and tools identified across CoreLogic
Business Criticality
80% of Applications are
Mission Critical or Important
Life Cycle Stage
48% of Applications in
Maintain or in Maintain
w/enhancements stage
Users & Use Cases
2M+ professional users
Sub-second to multi-day
transactions
Technology
63% of Applications utilize
JAVA (42%) or .NET (21%)
Mainframe
NATURAL
330 - Applications
Grow
Maintain
Enhance
3681 - Components
Physical
Logical
TBD
1009 - Data Stores
RDBMS
Object Store
Flat File
TBD
■ 532 - Tools
■ 300 - Externals
832 - Other
The CoreLogic Landscape
60. ■ Multitude of technology platforms
■ Complex, hard-wired, fragile and expensive
■ Today’s technologies are radically different from the past
♦ Mobility, Voice & Social Networks – Engagement norm
♦ “Platform as a Service” – Operating System norm
♦ “Infrastructure as a Service” – Compute & Processing norm
♦ “Data as a Service” – emerging ways of handling “big data”
♦ “Development as a Service” – Application build and deploy
■ Real opportunity to change what we do & how we do it
61
Fundamentals
61. CoreLogic Fabric
Engagement Services
CoreLogic Products & Solutions
CoreLogic Data Repository
Cloud Infrastructure as a Service
Ubiquitous Access
Single Sign-On
Solution Modules
(integrated or separate)
Re-usable Services
Scalable, Flexible & Efficient
Data & Analytics Delivery
Common Components & Services
62
CoreLogic Vision
63. 1. Developers focused on developing products, not managing tech stacks
2. Standard UI frameworks & style guidelines to speed up development
3. Components separated from applications allowing independent upgrading
4. Reusable services with built-in high availability, DR & elastic scalability
5. Resource flexibility enabled by standard technologies
64
CoreLogic Design Principles
64. Foundational Services
Platform Fabric
Infrastructure as a Service
Platform as a Service
Data as a Service
Development as a Service
CoreLogic Applications
Engagement Services
Products & Solutions
Web, Mobile, Voice Services
65
“Everything as a Service”
65. Development as a Service capability
Data as a Service capability
Scalability & Resiliency
Architectural agility
Hybrid infrastructure support
SDLC integration
Market adoption
Support & Operations
Ecosystem
Vendor lock-in concerns (Open)
Evaluation Criteria: Results of mini-POC plus following considerations:
Pivotal
Cloud Foundry
Google
App Engine
Salesforce
Force.com
Amazon
AWS
Oracle
Fusion
Red Hat
OpenShift
■ Conducted mini-POCs to assess capabilities
Engaging Technology Experts
66
66. Pivotal provides ‘open’ PaaS + Big Data Suite + Development Lab
Pivotal Partnership
LABS
■ Open Source Standard
■ Hybrid IaaS Support
■ Technically Sound
■ Industry Adoption
■ Extreme Agile
■ Pair Programming
■ Test Driven Development
■ Experience
■ State-of-the-Art Data & Analytics Tools
■ Strong Data Science Team
BIG DATA SUITE
67
68. 1. Exciting and rapidly changing times in the technology industry.
2. Capabilities available today weren’t around 2-3 years ago.
3. Enterprise organizations can now take advantage of the agility and
capabilities of Silicon Valley startups.
4. Companies who adopt this “new norm” have a competitive advantage
and can differentiate themselves in their markets.
69
Our Journey Begins…
69. Questions and Comments
Richard Leurig, Senior Vice President
CoreLogic Innovation Development Center
rleurig@corelogic.com
70
Good morning
Casey Hadden, developer, STO, SAS Institute
Walk through the case study, how we’re exploring
Where starting,
why CF of interest,
what we’ve done and learned,
peek at what’s next
SAS sells stuff
I’m talking about what’s next
We might not do that
So, nobody get the wrong impression
So, let’s start with where we are…
ISV
Customers are on premise
Mix of physical and virtual
Growing hosting and SaaS
70K customers and 150 products
Not just new, also moving current products and suites
Products have been around a while
SAS existed for 38 years
First was research project at NCSU on punch cards (300K LOC, 40 feet high!)
See that in CARDS; statement
Data to operate on
First platform entanglement
Punch cards go extinct and other platforms arise
Still have CARDS; today
But SAS had an opportunity to redefine itself on new platforms
So, we created: DATALINES;
Spirit of radical innovation and disruption to CF
In all seriousness
SAS has seen change 38 years.
Changes drive SAS to redefine
Originally, IBM assembler and PL/1 on mainframes.
Unix Workstations and PCs intro’d
Rewrite into C
two major codebases
C code backported to mainframe
IBM assembler retired
start of SAS’s MVA.
Client-server and basic web
SAS started Java for cross platform
Continuation of portability
web applications and server side
SAS embraced the middle tier concepts
Product list, bundling, and IT integration
Drive common components
Common needs to evolve separately
SOA concepts introduced
Today, all of that this basic picture of a SAS solution.
investigation of CF, 3 of the tiers here: web, services, and compute.
JVM,
heavy spring framework and related
run on the vFabric tc Server
common functionality provided by the SAS platform
security, configuration, and licensing.
All of that common, cross cutting stuff.
Solutions deploy alongside
build their own services and interfaces
Both sit on top of the SAS compute tier and analytics that drive SAS value.
servers are mostly C & C++.
Investigate both middle tier apps port
And how to best give those access to analytics
Move on to purpose – why, what drove?
STO ties business drivers and quality attributes to R&D work
CF obvious: cloud
Quality attributes like scalability, availability, reliability
But a more hidden one…
developers, developer enablement, developer experience.
SAS started as language
Language was interface for many years
Evolve with modern user interfaces
Core of analytics value is still in SAS language and servers
Solutions generate SAS code
today, developers are a force multiplier.
Great to get results without writing code
Value truly shows when embedded in processes
Where SAS user interface isn’t
Past wrote SAS code
But need access analytics other platforms
Drive for APIs and development platforms like CF
Tech office wants hoisting, rising tide
Logon application + SSO providers, Web application theming, server connections
CF has same qualities: scalability + reliability
main arch hoisting – MVA
Core of success – portability
Bulk is portable, small set if specific
Servers move to new platforms without a lot of work
But, that story is changing…
Previous OS of choice
Now IaaS of choice
Where CF and BOSH dramatically hoist applications
Large dev community
Many here for 20+ years
call introduction anecdote
Been through previous iterations
Know effort and leery
compare old and new
See CF architecture
And overlap SAS bits
Lots of commonality
SAS logon vs CF UAA
CF DEA pool vs SAS tc servers
Start from commonality instead of completely new
Have baseline and purpose, move on to actual investigation.
Aware of CF for a while
started looking more in January
Let’s see what has happened
Same architecture picture
Actually dealing with VA
Define VA
Focus on middle tier
Readily supported with java buildpack
Uses bits like Spring that are well known and supported
Lots of support for initial effort
Still need to connect to compute
Use user provided services to focus
Once we get one thing, another project for brokers
8 applications – services and UI
Changes related to tc Server ownership
Different living by yourself
Dependencies, JNDI, assumption: servlet context
Pull inward, self reliant, 12 factor
Other benefits - contention
Repetition
Move jars, change config
1 person, couple of weeks
Learn even without CF
Still want architectural qualities
Have concrete implementation to consider pros and cons
SAS Studio was next
Define SAS Studio
Different from typical solutions: scale up and down
Same JVM and analytic server components, but fewer when scaled down
Focus on CF integration
Service broker instead of UPS
single web application - services and UI.
same types of changes as before.
New bit is service broker for workspace
Define workspace server
Java + Spring is core,
Steve Greenberg skeleton
1 person, 1 week
Pivotal used to demo at SGF
Talked about specifics
Zoom out to larger takeaways
Here be dragons
Applications too big, long startup, memory settings
All telling us something
CF forces or lose benefits
Customize install for upload size, scaling, fork buildpack
Already doing that, so we don’t get as big a win
Hoisting helps, process falls into place
Single connection mechanisms let that be rewritten
Then spring let us easily switch
Then spring profiles let both side by side
Since we have drivers and attributes, CF can be enforcer
Encourages behavior we want anyway
CF citizens have good qualities
Smaller, focused, start quickly, non-fragile
Self sufficient, perform interrogation and initialization themselves
CF rewards that behavior making wins
Internal enforcer feedback
Now that we’ve gotten an idea about what it takes, where does that leave us to go next?
near term, still investigating.
Ported several major applications
have to apply across products.
Socialize across R&D
Tech office does in small groups
And large with tech radar
Have to make a decision
Are we going to pursue?
Another platform, preferred platform
THE platform?
Big puzzle is analytic servers
Have to make those readily available
Give positive deicision, several points
SAS Cloud analytics
Our cloud, SaaS
Private analytics cloud – your cloud or public
Not trivial
Can Cloud Foundry enable those deliveries?
Once running, enable developers
Drive up value
What analytical services to make available?
Forecasting, data quality, sas language execution
Finally, can we ship our products that deploy within CF?
Take CI and make the servers, services, and apps simply available?
The SAS case study
Personally, I think there is a lot of value here for SAS
Hope we come back and change might to did.
Thanks for your time
I hope you all enjoy the rest of the conference.