{"id":12404,"date":"2014-03-03T16:53:57","date_gmt":"2014-03-03T15:53:57","guid":{"rendered":"http:\/\/timoelliott.com\/blog\/?p=6495"},"modified":"2021-08-18T17:34:27","modified_gmt":"2021-08-18T15:34:27","slug":"agile-analytics","status":"publish","type":"post","link":"https:\/\/timoelliott.com\/blog\/2014\/03\/agile-analytics.html","title":{"rendered":"Agile Analytics"},"content":{"rendered":"

One of the biggest and frustrating barriers to successful analytics projects is lack of agility.<\/p>\n

Analytics systems must be built for constant change. The whole point of analytics is “actionable insight.\u201d The information accessed is supposed to result in changes to business process, and this in turn should trigger new opportunities and new analysis. \u00a0Business intelligence is like a bicycle \u2014 if it\u2019s not moving, it\u2019s falling over.<\/p>\n

And riding a bicycle takes a certain amount of agility. Analytics systems have been complex and slow while business change has been accelerating. In order to provide the core figures required to run the business, IT organizations felt that they had to protect the data warehouse from new or unverified sources.<\/p>\n

It isn\u2019t that IT doesn’t know how to do analytics \u201cright\u201d \u2014 they just couldn\u2019t act fast enough to satisfy their business customers. The result was regular crashes each time the business went around a tight corner and the analytics bicycle couldn\u2019t turn in time.<\/p>\n

The good news is that analytic technology has made huge leaps forward. The biggest benefit of new in-memory systems<\/a> isn\u2019t speed, it\u2019s simplicity. The data is stored just once, for both transactions and analytics (Gartner calls this Hybrid Transaction\/Analytical Processing or \u201cHTAP<\/a>\u201d). And there\u2019s no longer any need for “materialized views\u201d and \u201caggregate tables\u201d to ensure acceptable response times. This has eliminated most of the overhead and time required to prepare reliable data to answer business questions.<\/p>\n

In addition, systems like Hadoop have enabled the storage and access of “schemaless” information that doesn’t require the rigid, predefined structures associated with today’s data warehouses.<\/p>\n

Now that the technology has changed, analytic processes and organizations also have to adapt, and we\u2019re starting to see this process in leading edge organizations. Analytics competency centers, aware that they no longer have a veto over departmental IT investments, are taking steps to ensure that \u201cself-service BI” doesn\u2019t turn into \u201cself-centered BI.”<\/p>\n

At the last SAP TechEd Conference, I had the opportunity to host a session talking about Agile Analytics with my guests R<\/a>onald Konijnenburg<\/a>, SAP Mentor and Principal BI Consultant for CGI and Andy Steer<\/a>, UK Group CTO for itelligence.<\/p>\n

We had a great discussion covering topics such as:<\/p>\n