We are all now pretty much in agreement (although some of us more enthusiastically than others). It's not IF we're moving to the cloud, it's WHEN.
"So," you're probably asking, "How do I implement infrastructure, platform and applications in this brave new cloud world without messing it all up?" In other words, "How do I spin this cloud thing to get a promotion instead of a demotion?!"
Based on my experience, there are three big ways you can make a cloud implementation fail…and there are specific steps you can take to make sure you succeed!
How do I spin this cloud thing to get a promotion instead of a demotion?!
Fail #1: Inadequately securing your data.
According to the MIT Technology Review in the January 2, 2018, article, "Six Cyber Threats to Really Worry About in 2018," the number one cyber threat to worry about is more data breaches like the 2017 attack on Equifax. Even if your organization doesn't handle data as juicy as consumer credit reports, there is still plenty of content in your cloud solutions that might be of interest to hackers.
Want to avoid failure? Make sure to ask some key security-related questions as you evaluate cloud vendors, such as these questions shared by NetworkWorld.com
- Which of the vendor's employees has root and database access, and will anything prevent them from getting access to your corporate data? What controls are in place?
- Is data held encrypted? How?
- Is the held data separated between clients, or is it all stored on one huge database out there? How is data separated?
- How will the legal question of e-discovery be addressed should it arise as a business concern?
- Is the data flowing between the business and the vendor's cloud-computing infrastructure secured in some way?
- What controls would prevent vendor insiders from downloading your data onto a USB stick and walking out the door?
- In terms of service availability, can you get your vendor to sign a service-level agreement (SLA)?
- Is the vendor's data center in a location prone to hurricanes or earthquakes? What are their back-up plans?
- What information is captured in audit logs?
- Are there ways to limit where cloud vendors go within the corporate network?
Fail #2: Assuming that data sourcing, integration and storage will work the same way as behind your own firewall.
Data integration used to mean what you used for ETL (or ELT) tools, maybe with a data governance and dictionary solution thrown in for good measure. Buying these solutions was often like calling a plumber – they were a necessary part of the architecture but not something that was "front and center" in terms of the shiny, cool fun part of what we thought of in the overall blueprint. Not anymore.
If you want to get with the times, technology research firm Gartner is now calling this category "Enterprise Integration Platform as a Service." The old-school, on-premises data integration vendors don't even show up in that Gartner magic quadrant! So, Toto, as Dorothy said, "We aren't in Kansas anymore."
What does this mean in simple techno-speak? Basically, these vendors now provide support for bridging between a variety of technology protocols and data or message delivery styles and they supply agents, adapters and prepackaged integration flows to make cloud and on-premises integration work.
How not to fail? Make sure that data sourcing, integration and storage vendors are a central (and possibly primary) part of your cloud strategy. Don't move stuff to the cloud until you have very clear requirements and design for this part of your cloud initiative.
Fail #3: Presuming that "customization" means the same thing on-premises and in the cloud.
Customization in the halcyon days of on-premises environments was fun. You'd buy a Toyota and do requirements and design and end up with a Land Rover. In other words, you could completely hack a totally different solution within your on-premises solution. Frankly, the whole benefit of the cloud is that you aren't doing this; in return, you get automatic patches and upgrades and 99.99999% uptime!
Preventing failure here is very simple. Just remember one word: configuration. Don't try to do drastic or extreme things because you're a unique snowflake and your business model is very different than everyone else's. Accept that you get a pretty much 90% delivered solution with some options to modify settings and rules (kind of like an iPhone app).
EPM/BI in the Cloud? Want to learn more?
The OAUG has formed an EPM and BI Task Force that recently surveyed the OAUG's membership in this area to learn what they're doing in the cloud and how they're addressing some of these challenges. Find out where your peers are in this evolution and what steps the OAUG is taking to provide additional education and content in this arena in "EPM and BI Task Force Details, Survey Results and COLLABORATE 18 News!" (PDF file) from the spring 2018 OAUG Insight magazine.
About the Author
Kirby Lunger is a Partner at Performance Architects. She focuses on the Oracle enterprise performance management (EPM) and analytics arenas and co-chairs the OAUG's Analytics, BI and Big Data Special Interest Group (SIG).