Big bang approach to data cloud migration doesn’t work
Enterprises are learning that a ‘gung-ho’ approach with the intention to bulk-move quickly to the cloud, with no back-up plan, simply doesn’t work.
While there are definite business benefits to migrating data, there must be absolute clarity about the reasons to migrate, what solutions are available and the overall business focus.
Speaking at the 2020 ITWeb Business Intelligence Summit in Sandton, Johannesburg, today, Deepesh Thomas, head of Digital, Wealth at Standard Bank, said there are several advantages associated with the cloud – including elasticity to enable data analytics, automatic software updates and cost effective scalability. But, while it may be ‘the in thing’, it may not necessarily be custom fit for all organisations or necessarily the right decision.
“What we are seeing industry wide is people just flocking to the cloud without understanding what it means,” said Thomas. “Cloud has multiple advantages, however the way you adopt cloud as well as the solutions you put in place and the focus you’ve got is very important.”
Globally there is increased interest in going the multi-cloud approach. In Africa, there is growing interest in the hybrid cloud model because regulation within some countries does not allow for data storage outside the borders of the country.
Thomas said research shows that 84% of enterprises are adopting a multi-cloud strategy, driven largely by regulation and the desire to pass on the responsibilities of physically managing a private cloud.
Moreover, organisations leverage almost five cloud services on average to take advantage of other benefits =, such as being able to consume code rather than building it, security and flexibility.
“This is quite a dangerous statistic in my opinion because what you often find in organisations is a lot of tactical kind of approaches to cloud,” said Thomas. “So one business unit says ‘ok I want to go to cloud and I am going to adopt Azure’ and another wants to adopt AWS… All of a sudden, you’ve got five to ten different cloud providers, the data is not integrated, solution designers are not talking to each other, costs are spiralling out of control and you don’t have economies of scale.”
Thomas said it is important to choose the cloud strategy upfront, understand vendors and get subscriptions that are consolidated for economies of scale.
Standard Bank’s lessons
Reflecting on Standard Bank’s own data migration journey, Thomas said the aim was to improve the industry average quote to sale ratio of 12% - for example, leads generation based on existing clients whereby they are ranked on the probability of purchasing short-term insurance; or rankingnew business leads as they come into the call centre to improve lead closing rates.
The bank’s approach was to specify the problem, clean the data, build and train the model and test the model.
However, while there was a definite business plan and strategy in place, there was no migration strategy and no ‘plan B’ in the event of failure or difficulty.
Thomas said Standard Bank learnt a few valuable lessons, including the need to understand the data structures in detail, have a ‘plan B’ in place – i.e prepare for data recovery and have a roll-back strategy in the event of data migration or import failure, and approach migration with caution.
“There is no need to rush migration - start slowly. It’s always easier to import data slowly than to keep going back to remediate and fix data over time.”