<rt id="aarcs"><meter id="aarcs"></meter></rt>
    <cite id="aarcs"></cite>
    <rt id="aarcs"><meter id="aarcs"></meter></rt>
    <rt id="aarcs"><optgroup id="aarcs"><strike id="aarcs"></strike></optgroup></rt>
    <rt id="aarcs"><nav id="aarcs"><button id="aarcs"></button></nav></rt>

    1. <rt id="aarcs"><nav id="aarcs"><p id="aarcs"></p></nav></rt>
    2. <rp id="aarcs"><nav id="aarcs"><option id="aarcs"></option></nav></rp>
      <tt id="aarcs"></tt>
    3. <cite id="aarcs"></cite>
      <cite id="aarcs"><noscript id="aarcs"><var id="aarcs"></var></noscript></cite>
    4. June 20, 2019

      Are We Ready For Autonomous Software Implementations?

      Dennis Jolluck

      Dennis Jolluck
      Vice President - Applications Development & Field Product Management - Latin America Division/Oracle

      Share This Post

      During a recent Oracle Open World, Larry Ellison introduced the first Autonomous Database based on the concept of “machine learning”, where computers read Tera-bytes and Zeta-bytes of data in order to find normal data patterns versus abnormal patterns. For example: Your CFO who is based out of Southern California (and not travelling) had just logged into your Financial Systems at 5:00am from Hong Kong through an unfamiliar IP address. This is the perfect example in trying to address cybersecurity and data theft through “machine learning”. 

       The Autonomous Database has to be able to “patch” itself while running, with no scheduled downtime and no pending management approvals.  100% automated:  automatically provisions; patches; updates; continually tunes and backs itself up without any human intervention.

       So what happens to the role of the DBA? Like everything else (as a result of technology) that has occurred over the last 15 - 20 years, the role and the responsibility of the DBA position will evolve. From routine backups and day to day administration to more focus to database design, schema design, developing data analytics in relation to “machine learning”, and “mission critical” disaster recovery.

       As I was listening to Larry’s presentation, I was correlating the Autonomous Database to Enterprise Application Implementations. Could this be the next “new wave” for significant improvement? Can we automate configurations with even 80%+ accuracy?

       First, let’s define a traditional application implementation scope for (e.g.) ERP:

      • An enterprise must hire a consulting team (3 – 5 consultants) at a substantial cost for anywhere from 4 – 9 months; the cost may vary but you are looking at a minimum investment of $200K. The average cost could easily be two to three times that amount.
      • The consultants’ domain expertise may vary by product line: e.g. General Ledger, Accounts Receivable, Accounts Payable, Technical expertise, etc.
      • The consulting team will communicate with the users and manually collect, define and document their ERP business requirements and compare them between the existing system and the proposed system.
      • The consultants will manually establish the business flows and set-up functionalities.
      • Mistakes will be made with those setups. These mistakes will vary depending upon the consultants’ degree of experience or inexperience or whether they even attended training. How expensive are those mistakes to the customer, to the project, to impacting the “go live” date?
      • Functional indecision and/or implementer mistakes will result in logging a service request to the vendor. More time wasted. More communication and clarification required with the software vendor. Additional impact to the “go live” date.
      • Essentially it will take at a minimum 2 weeks to load all of the setups and start the user acceptance testing. But the average time is more likely 4 – 8 weeks.

      ERP implementations are challenging and complex. Numerous failures have been reported over the years. If not managed properly, the above process can be costly and potentially detrimental to an enterprises business. Regardless, can we use todays’ technology and leverage “machine learning” and “intelligent applications” and automate an implementation? How close are we to addressing the above challenges and provide a fully automated injection service that loads pre-tested setups for a provisioned instance, with personalization in perhaps a few hours? If you break down the steps of this process, we could envision the following:

      • Collect the requirements based on answering a comprehensive online questionnaire.
      • Based on answering the questionnaire, artificial Intelligence takes over and automates the correct combination of setups with a high degree of accuracy.
      • Flexible and automated mapping for the Chart of Accounts, which is the “heart” of an ERP implementation.
      • Within 24 - 48 hours a fully configured instance is delivered to the customer.
      • As a result, the number of test cycles are drastically reduced, since the setups are “near” perfect and have been fully tested. Perfection minimizes the potential service requests logged to the software vendor.
      • Data acquisition is simplified. There is only one tool to learn to automate the loading of data.
      • Transition between Test & Production environments is fully automated.
      • Potentially an enterprise can “go live” within 6 to 8 weeks which includes:
        • Functional set-ups & business flows fully tested;
        • Full setup documentation including business flows and functionality;
        • A list of functional “gaps” and customizations from your existing system that are NOT addressable with the new system;
        • No software to install once the setup is completed;
        • Software Personalization is also available; and,
        • Potential replication for additional Legal Entities for a fraction of the cost and time.

       What are the implementation challenges today and the potential opportunities?

      • For the Customer:
        • Delayed “Go live” dates which results in additional cost in maintaining the current system.
        • Exceeding the projected budgeted costs for the entire project.
        • Unable to take advantage of the new systems anticipated benefits and technology.
        • Employees working two shifts supporting & testing two ERP systems.
      • For the Software vendor:
        • Receiving service requests from numerous implementations with the same problem.
        • Service requests are perceived by the customer as a complex product to implement or even unstable.
        • Customer satisfaction and referenceability gets off on the wrong foot; effort required to rebuild the confidence in the product and the vendor.
      • For the 3rd party Consulting Team:
        • Discover some of your consultants were not properly trained or skilled. May have to discount or provide “free” services to make amends to the customer.
        • How much time is wasted chasing a “red herring”? Is it a bug or a miscommunicated requirement? A function that was not configured properly? Or an issue with the tech stack?

      A Machine Learning Implementation tool for Enterprise Software could provide the following benefits:

      • Customer: The number one priority is the “go live” date is on time, on budget, and become instantly satisfied. They want to ensure they get all of their requirements and solution configured at the lowest cost, in the fastest time, with the highest quality. 
      • Software Vendor: With an “automated” implementation, less service requests to address; more attention and focus to manage a customer’s complex issues & proactive customer care (as compared to addressing issues related to “poor” implementation skills); improved margins and a high probability of customer referenceability.
      • 3rd party Consulting Team: Have the opportunity to manage double or triple the number of projects simultaneously with the same consulting headcount. As well, they have the occasion to transform their practices skillset from low-value “configuration consultants” to higher-value “Digital Strategy/Transformation consultants”.

      I can speak at least for Oracle Cloud implementations, these automated machine learning techniques are available and beyond the “proof of concept” stage.   However, the question for 3rd party implementers is:  Do they have the foresight to change their consulting business model or will they wait to be cannibalized like Amazon has done to practically every industry? 

      Share This Post