How development and humanitarian organization should approach cloud services.
Back in 2006 I was helping a UN organization develop its contingency planning for the avian flu (H1N1) in Laos. As any Country Office we had a server room, running AC 24/7, to operate in case of an epidemic the confinement time was 6 weeks. I remember the look of the Country Director when I told her that we had to store few hundred liters of fuel in the backyard of the office to run the servers!
Running servers on premises comes with a cost (set up, maintenance, etc.) that is less and less relevant for small to medium organizations.
Today, the offer of Cloud services keeps expanding and it does not take a lot of time to conduct a cost benefit analysis pointing towards cloud services. The main hurdle to shift from on-premises to cloud based infrastructure for many organizations in the humanitarian sector was security. But this concept that cloud is less secure than a non-cloud option is gradually dying especially when you put that in front of all the other advantages of cloud services (price, reliability, regional replication, performance, scalability and the list goes on). Even the Pentagon recently launched a tender to host its data on the cloud and if there is one thing I learned in my few years working for the French MoD is that data security is extremely sensitive.
Hosting its infrastructure on the cloud is today becoming an evident choice. Most of large private actors already have a cloud based architecture or at least a hybrid architecture (cloud and on-premises infrastructure) serving as the foundation of their digital transformation enabling their teams to access the best solutions to meet their unique
needs without arbitrary constraints that impede progress. It is evident that NGOs, UN and Red Cross movement organization should follow the trend.
But here is the catch. What to choose amongst the 1000s of available options? How shall I move forward? What’s my next step?
Nearly every day, each and every cloud provider is releasing a new product. It is quite a challenge to keep up with the speed and complexity of new products. It seems it is a new game of some provider to make cost estimates of their cloud services almost impossible, not because they hide the prices but because of the complexity of their system. You pay according to the number of emails sent, the size of attachment, the number of GB you store but pricing differ if you do not accessed them in the last month, the number of GB uploaded and downloaded, the size of server that in ant way auto scale over night, the replication, etc…
In addition, their platforms are more and more vendor specific, cornering certain organizations in a vendor lock-in where it becomes almost impossible to opt out. Some provider develops their own flavor of a well-known open source database that only operates on their infrastructure.
But with risk comes opportunity. Organizations have a tremendous opportunity to choose the exact product that match their needs at any time. You need to store high volume of data with infrequent access, one provider will have the best product, you need a top of the line cloud computing dispatch in several regions, another provider might have the solution.
The game today is not to decide or not to shift to cloud based infrastructure but to equip your organization with the ability to use different services at different times; moving to a multi-cloud platform where some databases might be hosted by one provider while others by another and so on.
There are several options to achieve that, leverage open source for interoperability, standardize governance across platforms and tools, centralize management, etc. But one thing is sure is that it will become more and more challenging for one organization to have the internal skillset to know what would be the most suitable set of options. Third-party support for cloud infrastructure design is becoming a full and independent technical field where integration and cross-platform interoperability and integration are the key technical fields.
One organization retaining such high level types of skill set might not be cost effective, but not relying on experts for cloud infrastructure will potentially lead to higher costs and weaken security and safety.