Enterprise SaaS Challenges

I’d like to share some of my thoughts about the challenges that Enterprises are facing in leveraging the SaaS service model.

As always, your comments are welcome!

Top challenges:

  1.   Data Access
  2.   Integration
  3.   Operations

Data Access

  1. Data model needs to be decoupled from the business logic and/or the analytical model:
    1. Flat data models (as in Non-Hierarchical) ALWAYS work
      • Dependencies and hierarchies inside data can be represented as relationships (or links, tags, etc.)
    2. Data transfer and sharing data becomes much easier
      • Moving data puts a lot less stress on the network, therefore improved data transfer performance
    3. Granting/revoking access to the data becomes less complex to manage
    4. Data protection and tampering detection become standard capabilities of the Data Layer
      • Because of its natively granular structure, compromised data can be easily quarantined and removed from the processing pipeline
  2. Business logic is much more fluid than the data itself:
    1. Once the data is gathered and stored there is not a whole lot going on at that level. The focus shifts from gathering data to processing
    2.  Any hierarchy (or multi dimensionality) inside the data can break the business logic to the extent that the business logic needs to take the data hierarchy (literally!) and put it together in a form that makes it easier for the business logic
    3. Usually the same piece of data is part of several very distinct processing pipelines


  1. Data integration
    1. Data portability (with respect to integration) is difficult to achieve
      • Enterprises have data silo-ed in data marts, warehouses, etc; integration becomes a custom act and there is very little leverage to it
    2. Coherent data security design is not trivial to implement
      • Each data source implements its own security design
      • An unified security design should NOT replace the security features built into each data store, it should ONLY leverage them instead
  2. Application integration
    1. Integration with non-SaaS (legacy) Enterprise applications
      • Enterprises need options for leveraging (sometimes very large!) investments made in legacy applications
    2. Integration with SaaS Enterprise applications, either Public or Private Cloud
      • Ideally, the integration needs to occur at the very top of the application stack, via APIs
      • Any lower level integration sooner or later becomes unsustainable; creates tight dependencies that are error-prone and difficult to maintain


  1. Full lifecycle
    1. Plan not only for the initial rollout but also for monitoring and reporting at runtime, integration, security and containment, self-service
    2. Data has an “interesting” characteristic of compounding organically over time; make sure that you plan for the right capacity
    3. BYOD is no longer the exception; make sure that users get the same level of service on their mobile devices just like on the desktops
  2. Costs
    1. From the very beginning understand exactly the cost structure and fit that into an ROI model
    2. Use the ROI indicator to ALWAYS measure and adjust costs