⇓ More from ICTworks

Cloud Computing for Development: A Four-Stage Analysis of Public, Private, and Hybrid Solutions

By Guest Writer on June 5, 2017

The idea of using the cloud for development is gaining popularity. In a 2013 assessment of “The Cloud Economy and Developing Countries” United Nations Secretary-General BAN Ki-moon lauded the cloud’s potential for fostering economic and social development and furthering Millennium Development Goals.

A 2016 World Economic Forum report said with help from cloud provided services, “… the financial barriers to new business entry have fallen significantly,” providing both startups and small to medium enterprises the ability to scale and remain resilient.

Once an organization decides to switch from its dog-eared, paper-based field data collection system to mobile devices, it should ask itself two questions:

  • Where will the data be stored and what level of data analysis needs to be done?
  • What type of backend storage and computing system best supports our organization’s work?

One Size Cloud Does Not Fit All

There are three main types of “cloud” data storage, public, private, and hybrid.

You can use a “public cloud” server like Amazon AWS, Microsoft, or Salesforce.com (for which our product TaroWorks is a mobile data collection and analysis, offline CRM and field force management tool). In all of these cases, the same set of data and computing applications are stored on the cloud provider’s servers and are accessible to multiple users via the internet.

Or, you can employ an on-premises solution (also know as a “private cloud ”), which stores data and applications locally on an organization’s own servers and makes it accessible to multiple users through internal networks – over which you have more control.

There are benefits and tradeoffs to each, not to mention a hybrid cloud option that combines the two, but more on that later.

4 Decision Points for Cloud Computing Services

In the less lofty world where social enterprises and NGOs make hard choices about field data collection and analysis (as well as data storage) and need technology platforms to manage sales teams and supply chains, the tradeoffs between cloud, on-premises or hybrid options often turn on:

  • Cost: Can you afford to purchase hardware and software? Do you have the in-house expertise to manage servers or build applications?
  • Connectivity: What are the mobile data and fixed line bandwidth constraints on the ground?
  • Collaboration: How important is it for your team to collaborate in real-time with far-off colleagues, customers, beneficiaries or donors?
  • Data Complexity: How much data do you need to collect and store? How complex is the data analysis you’re doing? Would you have to migrate existing data from legacy systems to the cloud? Does your data include personal health information or government owned data, both of which require top levels of privacy and security?

Cloud Computing Analysis

Here’s how the cloud, on-premises and hybrid cloud stack up when measured against the four factors – cost, connectivity, collaboration and data complexity.


  • Winner – Public Cloud: SMEs and nonprofits can reduce upfront and ongoing hardware and software costs by utilizing cloud hosting as it relies on servers and system architecture owned and managed remotely by Infrastructure as a Service (IaaS) or Platform As A Service (PaaS) cloud providers as well as software applications and mobile apps developed, hosted and regularly enhanced by Software As A Service (SaaS) cloud companies like TaroWorks.The on-premises computing and data storage approach, on the other hand, requires expertise in server setup and maintenance, security and in some cases application development. Software Advice, a Gartner company, has a calculator, which helps weigh the costs of a cloud system versus an on-premises setup.


  • Winner – Hybrid Cloud: There’s an obvious hurdle to using the cloud in developing countries when you’re working outside of cities. Your ability to connect to remotely hosted databases from the field or use cloud driven CRMs is only as good as your internet or mobile data connection. If you need to access data directly from a location where fixed and mobile data infrastructure and bandwidth is spotty, one alternative is building an on-premises data storage system that hosts data in a Network Attached Storage device and connects to users over an internal network.You’re still without internet connectivity, though, which is where hybrid clouds can give multiple users access to both locally stored content and periodic connections to the internet for downloads. Brck’s Kio Kit is an example of this hybrid cloud approach at work in Kenyan schools and its recently released SupaBrck is a “solar-powered Wi-Fi box that operates as a 3G hotspot and off-grid server.”

    Operating in areas with little or no internet or mobile connectivity is not the death knell for accessing the cloud from a mobile device either. A growing number of mobile apps like TaroWorks or custom built apps using Open Data Kit  allow field teams to collect data offline and in some cases, navigate geography or even view video with no connectivity. The newly collected information saved on a smartphone or tablet can be synced back to the cloud when within range of mobile data or internet service.

    We’ve focused on mobile data collection and field operations management in the African, Latin American and Asian countries where fixed (Ethernet) and mobile bandwidth are becoming more robust, according to Cisco’s Cloud Readiness Index.


  • Winner – Public Cloud: You’ve collected the data – now what? Having a web-based destination accessible by multiple, simultaneous users (from different time zones) where data can be visualized, shared and discussed in real-time, is one of the cloud’s strengths.Using the cloud can bust through data silos  – created when program data is housed on local servers in each country where it is collected –  by letting anyone in an organization view the same data regardless of their location. It’s also a way to provide access to metrics for customers, beneficiaries, donors or business partners. Connected to a smartphone or tablet, cloud-hosted CRMs are able to push survey results or content back out to the field.

    Our customer BOMA Project, which implements a poverty graduation program in rural Northern Kenya, uses mobile devices and cloud-hosted data analysis to close feedback loops between the women entrepreneurs it supports and the experts providing them with advice to launch small businesses and start a savings program in their community.Data on business progress and feedback from each entrepreneur can be uploaded and analyzed quickly allowing BOMA business coaches to adjust the advice they give these women based on data results and trends.

Data Complexity

  • Winner Scalability – Public Cloud: As an operation’s data and computing needs grow, cloud systems expand quickly and for less upfront hardware and software cost than on-premises computing. Depending on the amount of data to collect and the complexity of the data management and analysis, the cloud can also apply more computing power to problem solving than you’d be able to on your own.Our customer iDE employed the cloud to balance supply and demand in its Cambodia-based latrine manufacturing and sales WASH program.Using mobile data collection and cloud-based CRM tools, iDE tracked goods from factory to installation. It counted on the cloud’s flexibility to quickly meet a need for more field data collectors and data storage as the program ramped up. The iDE team now manages over 6,000 orders for sanitary toilets a month, operating a complex network of 200 sales agents, 100 independent latrine manufacturers and 25 supply chain coordinators.
  • Winner Data Migration – Public Cloud: Moving locally stored data and related applications to a cloud server is one of the most daunting tasks when switching from on-premises to cloud. It can even be complicated transferring content between two cloud-hosted databases. Don’t underestimate the challenge and rank the cloud service provider’s data migration capabilities among the more important selection criteria.The good news is there are subscription services you can purchase to help automate the flow of data between two technologies and data integration consultants who can help make the transfer successful – frequently by using an API.
  • Winner Data Security – Private On-Premises: The spate of news reports about database hacking, malware, phishing campaigns and denial-of-service attacks have heightened security concerns about hosting data on servers accessible through the internet – over which the cloud user has no control.That has prompted some organizations, especially those collecting sensitive health and government information, to opt for an on-premises database that can be placed behind a firewall they manage, subject to permissions they control, using infrastructure they’ve configured. The large cloud service providers take pains to bolster their data and hardware security so check their trust and security web page to see if it gives you a level of comfort.If that’s not enough, there are tools that can help verify and strengthen the integrity of all forms of cloud hosting.

    A study of SMEs in Nigeria found that these businesses actually value the cloud as a more reliable data storage option than on-premises alternatives when confronted with data storage risks like frequent loss of power.

  • Winner Data Sovereignty – Private On-Premises: If you are working with government data, hosting that information on servers outside of the country from which the data originates can be illegal. In Nigeria, for example, there are no comprehensive cloud usage regulations but all government data must be hosted in-country. (Read More)On-premises computing allows you to keep the data in the country of origin as does a local cloud service provider. Check on data sovereignty rules for the countries in which you’ll be collecting information and review the data hosting policy of any potential cloud hosting platform.

Has using the cloud for ICT4D achieved the promise hoped for in the United Nation’s 2013 assessment? No, but it has become a serious option for data collection and analysis as well as managing field operations. The challenge is determining whether what the cloud does well is what your organization needs most.

By Kay Chau, Director of Product and Customer Success at TaroWorks

Filed Under: Data
More About: , , , , , ,

Written by
This Guest Post is an ICTworks community knowledge-sharing effort. We actively solicit original content and search for and re-publish quality ICT-related posts we find online. Please suggest a post (even your own) to add to our collective insight.
Stay Current with ICTworksGet Regular Updates via Email

9 Comments to “Cloud Computing for Development: A Four-Stage Analysis of Public, Private, and Hybrid Solutions”

  1. Faizan says:

    Thanks for this write-up Kay. I think outlining the different hosting options – cloud-hosted, on-premise, hybrid – is useful for users.

    I think one of the biggest considerations – if not THE biggest consideration – that we unfortunately don’t see given enough weight in the ICT4D community (even in your post, it got secondary billing) is how to ensure your data that is hosted in the cloud is kept secure. Whether one is using software hosted on a “public cloud” or one has an on-premise/”private cloud” server that is connected to the internet, the first thing users should think about is – in what ways is data stored on such a server vulnerable? And how is my data being protected?

    In fact, it is a bit misleading to say that on-premise hosted solutions are more secure. From a technical perspective, a server in a data center that is in the US (say, an AWS data center) and plugged into the internet or a server that is in your local office and plugged into the internet are no different. The distinction between “private cloud” and “public cloud” is really the sovereignty argument that you also mention: who owns the server? But as long as your server that is storing your data is connected to the internet, whether on-premise or a public cloud server, it is equally vulnerable.

    So the bigger security consideration is whether the software you are using (whether on a public or private cloud) allows your data to be encrypted at-rest. That is, is the data encrypted and unreadable while it is sitting on the server? That is the only way to truly secure your data as that ensures that not just anyone who could access your server (such as the vendor who provides the software you are using!) can view the data. Unfortunately – and most vendors will never admit this – while most tools encrypt data “in transit” (while the data is traveling from, say, a tablet in the field to your server), almost no tool out there encrypts data at-rest, while it is sitting on the server. And so I would argue that probably 95% of data being collected today is woefully insecure, no matter where it is being stored and what tool is being used to collect and store the data.

    If the tool you are using can visualize/display your data in the cloud, the tool can only do that by being able to read the data, which means the data is also readable by anyone with access to that server (including the vendor and its employees). For example, data hosted in Salesforce is not encrypted at rest – so the engineers and system admins at Salesforce can read all that data if they wanted to (and companies can have internal policies about not doing that, but talk is cheap and that is no guarantee of security). There are exceptions (at SurveyCTO we worked with data security researchers at Harvard to allow users to encrypt data at-rest while still being able to visualize and monitor incoming data) but very few tools even allow at-rest encryption, let alone make users aware such an option exists and that they should be using it.

    When users do decide to use software that involves collecting or storing sensitive data – whether they prefer a public cloud or private cloud – they should make sure that the software they are using has easy to use at-rest encryption features, to enable them to properly protect their data.

    Also, at-rest encryption aside, a private cloud is, in practice, less secure than a public cloud. This is because most organizations that decide to go with on-premise hosting do not have the technical resources or know-how to properly secure their “private cloud” server. For example, to ensure that your on-premise server has the same level of firewall protections and security that comes with an AWS hosted server would cost an organization in the tens, if not hundreds, of thousands of dollars a year. So most on-premise servers end up, in practice, being poorly protected and maintained. Using software hosted in a reliable public cloud coupled with using at-rest encryption is usually the most secure set of options available.

    • Kay Chau says:

      Faizan: Thanks for your comments! Great insights – the good news is Salesforce does in fact encrypt data “at rest” so that’s not a problem. Here’s more information if you’d like the details.



      • Wayan Vota says:

        Kay, I would say that Fizan points to a problem not simply fixed by saying Salesforce stores data w/ encryption. Salesforce is a “public” cloud solution and doesn’t help the very people he points out as most vulnerable – orgs who self-host their data.

        I agree – the next major development scandal will be data loss.

        • Kay Chau says:

          Wayan, thanks for your comments and for publishing TaroWorks’ post on ICTWorks. We’re definitely kindred spirits when it comes to concerns about data security risks and the need for even more diligence and training if organizations decide to host data in the private cloud. My reference to Salesforce wasn’t intended to present encrypting server data at-rest as a solution to private cloud hosting security challenges but rather to correct the statement that Salesforce did not have at-rest encryption capability in the first place.

          Data loss threats (in private and public cloud hosting) keep me up as well and I would add to that, the challenge of managing data sovereignty issues as cloud use expands in developing countries.


  2. Herman Fung says:

    Thanks Kay. Really good in depth analysis and discussion points. Love the Software TCO Calculator, which I wasn’t aware of before.

  3. Sonali batra says:

    hi Kay. Thanks for the write up. We at Operation ASHA are using the public cloud (AWS) to store patient’s fingerprint scans and we have found that it works well for us. Operation ASHA is the largest TB treating NGO in India and third largest in the world in terms of patients treated. We connect the last mile by providing medications and other services to the doorsteps of the slum dwellers. We have a biometrics based eCompliance software that helps us ensure that patients are compliant to the long regime of DOTS treatment. We use an Android tablet connected to a fingerprint reader/iris scanner. When a patient comes into the program for the first time, she is enrolled into the system by taking finger scans/iris scans and capturing demographic details like name, age, address etc. Subsequently when the patient comes to the center to take her medicine, her fingerprint is scanner along with the fingerprint of the health worker, This proves that the dose has been taken under direct supervision. If a patient does not come in when supposed to, the system generates a text alert and sends it to the health worker who then goes to the patients house to give her the medication.
    The tablet can work in offline mode and is synced to the AWS server every 36 hours. The data on the server is then analyzed and visualized using our in house built reporting system.
    Biometric data ensures that results are accurate and that there is no data fudging.

    • Howard Sherman says:

      Sonali: Thanks for sharing this really interesting application of cloud computing. Two questions if you don’t mind:

      *Is the biometrics software you’re using on your tablets proprietary and custom developed in any way for you or is it an app that can be downloaded from Google Play and configured for your specific use? If the later, what is the name of the app? I’m interested in its other possible applications.

      *What are some of the challenges you’ve faced using cloud computing in the last mile?

      Thanks again!

      • Sonali batra says:

        hi Howard,

        The Biometrics software that we are using is proprietary though it can be downloaded from the Play Store as in it needs an authentication key to work.

        Regarding challenges, I think the only challenge we faced so far was the migration of our server and database to the cloud.

        Please let me know if you have any other questions,