Monday's Musings: SaaS, SOA, Integration and How To Make A Peanut Butter And Jelly Sandwich In The Cloud


Rapid SaaS Adoption Will Lead To A Repeat Of 1990's Best Of Breed Integration Challenges

The proliferation and rapid adoption of SaaS solutions stems from 7 key benefits: richer user experience, rapid implementation, frequent cycles of innovation, minimal upgrade hassles, always on deployment, subscription pricing, and scalability (see Figure 1). Despite these benefits, organizations head full circle towards the same best of breed dilemma they faced in the late 1990's.  In that era, organizations sought innovation from more nimble and agile competitors.  The result - a concerted effort to deploy a number of on-premise, point solutions.  Willing to sacrifice not having a single instance for functionality, they invested heavily in integration.  Almost a decade later, organizations will encounter similar challenges with harmonizing a plethora of SaaS entry points in the next 2 to 3 years. Given the growing number of SaaS solutions at cost-effective price points and easy adoption, today's organizations face problems in a geometrically larger scale.
Figure 1. Seven Benefits of SaaS Deployments
screen-shot-2009-11-08-at-22936-pm
Modeling How To Make A Peanut Butter And Jelly Provides Key Insights Into The Integration Challenge
Today's integration challenges move beyond data integration to include process level and meta-data requirements that span across a range of business processes and relevant key performance indicators (KPI's).  As more solutions are added, organizations will want to model their end to end business processes as web services and support synchronous and asynchronous communication protocols across hybrid deployments.   Organizations can expect canonical data models play a key role in harmonizing business objects.  To put this in real world terms, imagine describing how to make a peanut butter and jelly sandwich using a hodgepodge of solutions.  Let's take a look:

Example 1:  Modeling in a .NET application

  1. Bread: take 2 slices of bread
  2. Peanut butter: spread peanut butter on one slice
  3. Jelly: spread jelly on the other slice
  4. Assembly: put the bread together
  5. Assembly: slice down the middle
  6. Delivery: serve on plate

Example 2:  Modeling in Force.com

  1. Bread: take 2 slices of bread
  2. Bread: determine whether or not to toast the bread
  3. Peanut butter: choose chunky or creamy
  4. Peanut butter: spread peanut butter on one slice
  5. Jelly: choose type of jelly
  6. Jelly: spread jelly on the other slice
  7. Assembly: put the bread together
  8. Assembly: determine if the slices is in half or diagonal
  9. Assembly: slice down the middle
  10. Delivery: choose type of plate (e.g. paper or plastic)
  11. Deliver: serve on plate

Example 3:  Modeling in NetWeaver

  1. Bread: take 2 slices of bread
  2. Bread: determine if the bread is organic or not
  3. Bread: determine whether or not to toast the bread
  4. Bread: determine how light or dark the bread should be toasted
  5. Peanut butter: determine if the peanut butter is organic or not
  6. Peanut butter: choose chunky or creamy
  7. Peanut butter: spread peanut butter on one slice
  8. Peanut butter: determine thickness of spread
  9. Jelly: choose type of jelly
  10. Jelly: determine if the jelly is organic or not
  11. Jelly: spread jelly on one slice
  12. Jelly: determine thickness of spread
  13. Assembly: put the bread together
  14. Assembly: determine whether you want the crust or not
  15. Assembly: determine how to slice the bread (e.g. diagonal, half, 4 cubes, etc.)
  16. Delivery: choose type of plate (e.g. paper or plastic)
  17. Delivery: determine garnishes with the sandwich
  18. Delivery: serve on plate

In these examples, notice how they granularity of processes become deeper and deeper within more complex solutions.  How would you take the peanut butter web service from the .NET example and harmonize this with the NetWeaver example?  Now take this real-life example at a hypothetical global pharma:

  • SAP financials (on-premise)
  • Oracle JD Edwards manufacturing (on-premise)
  • Salesforce.com CRM (SaaS)
  • Workday HR and Payroll (SaaS)
  • Concur Expense Management (SaaS)
  • Xactly Incentive Comp (SaaS)
  • NetSuite OpenAir Project Management (SaaS)
  • Ariba Spend Management (SaaS)
  • Gmail and Google Docs(SaaS)
  • Jive Community Platforms (SaaS)
  • SocialText (SaaS)
  • WebEx (SaaS)

Recommendations
As organizations consider SaaS adoption they must put into place an integration framework to support the competing forces of innovation and harmonization.  These integration frameworks must consider not only data, but also process, metadata, and business intelligence.  Key suggestions include:

  • Begin with the end in mind. Identify the key performance indicators.  Determine how to measure business value
  • Understand your key business processes. Classify your business processes into 3 buckets: commoditized, mission critical, and innovative.  This way you'll know which processes can be put into an outsource, shared service, or internal ownership.
  • Map the granularity of the business processes.  Group similar processes across different solutions and understand the levels of granularity.  Identify points for harmonization.
  • Determine the data integration requirements. Identify the key business objects associated with the business process.  Ensure that the right data arrives to the right process at the right time for the right person.  Map key meta data to process and business objects.  Build out your canonical data models.
  • Build loose frameworks for evaluation of SaaS solutions. Give line of business teams guidelines to determine how SaaS solutions fit into existing processes.  Use this to jump start integration and proactively identify integration challenges.
  • Determine approach and SaaS adoption policies. In some cases, point to point will make more sense. In others, greater levels of integration and control may be required.  Avoid a one-size fits all methodology in setting up policies.  Consider the business case first and foremost.

The Bottom Line - SOA's Not Dead And Integration Is Key To Successful Hybrid Deployments
Given these scenarios, CIO's and line of business apps will need to rely on stronger enterprise architecture and integration in hybrid deployments.  In fact, au contraire on the death of SOA!  Introduction of next generation social enterprise apps will only accelerate the need for good architecture and services design. Expect solutions from Boomi, Cast Iron, Informatica, Pervasive, SnapLogic, and Talend to play a key role going forward.

Your POV

Where are you with your SaaS deployment strategy?  Have you considered SaaS integration tools? What are you using and why?  Do these issues resonate with you?   Who owns the larger integration problem in your organization? Let us know how we can assist or please post or send on your comments to rwang0 (at) gmail (dot) com or r (at) altimetergroup (dot) com and we’ll keep your anonymity.
Copyright © 2009 R Wang & Insider Associates, LLC. All rights reserved.