EDITOR ’ S QUESTION

KUNAL AGARWAL , CEO , UNRAVEL DATA

Governance controls need to be implemented and maintained at the technical level in order to meet a wide-variety of concerns .

It interfaces nicely with governance and security concerns , so we discuss these issues with customers all the time .

So , this question about cloud access is valuable but it ’ s tactical . It ’ s part of a bigger question : how can IT leaders have better visibility over access to all of an organisation ’ s data processing tools and all of its data , whether the code and data involved are on-premises , in any cloud , or in transit .

Metadata leaks , at very large scale , have been hugely embarrassing and costly to some of the companies that have suffered them .

Cost concerns also come up forcefully in the cloud . On-premises , costs are largely sunk – you paid for a certain number of servers , within a deliberative process and any large expenditure is closely scrutinised .

The cloud is ‘ pay as you go ’; if you turn that around , it means , ‘ as you go , you pay ’. When you ’ re working at scale , it ’ s easy to run up a six-figure bill in the cloud in just a couple of days , even if you ’ re just testing a new service on production-type volumes of data .

We have careful performance and resource use tracking in the Unravel data platform . As people have moved to cloud , we are adding in explicit cost reporting and controls . You can put a ‘ stop loss ’ on cloud spending for a job – for example , alerting you , or even pausing a job if spending on a workload reaches , say , four figures in cost .

You should always have access to your current and budgeted spending for all your data services . We think of this as a DataOps issue .

DataOps is the intersection between data engineering , data processing and operations . It ’ s like DevOps but it starts with data flows first .

When software is in development and being run against test data , or copies of existing data , it can more easily be kept controlled . But when software and data are used in production , you get operational concerns .

These questions gain a sharper edge in the cloud because it ’ s new , so let me answer this directly . Governance controls need to be implemented and maintained at the technical level in order to meet a wide-variety of concerns , including security of code , data and metadata .

So , it may be that moving some workloads to the cloud is exposing the fact that you didn ’ t have the controls needed – which , perhaps you should have had – onpremises , as well . Take a holistic approach to your entire data and processing estate .

‘ Solve the problem ’ across your organisation , rather than playing ‘ whack-a-mole ’ as questions get asked , or problems arise . You ’ ll sleep better at night and so will your customers and other stakeholders .

You should always have access to your current and budgeted spending for all your data services .

36 INTELLIGENTCIO EUROPE www . intelligentcio . com