Hi Myrle, First of all, my sincere apologies for this crazy slow response. I somehow overlooked your email/posting. Sorry about that! I will pay better attention moving forward.
Thank you for all of your thoughtful questions. I'll do my best to provide some answers inline below. -----Original Message----- From: Myrle Krantz [mailto:[email protected]] Sent: Tuesday, August 1, 2017 7:32 AM To: [email protected]; [email protected] Subject: Re: Gen 3 and Apache Fineract On Fri, Jul 28, 2017 at 5:24 AM, Henrik Esbensen <[email protected]> wrote: > Hello Apache Fineract community, Hi Henrik. You and I are a little better acquainted, than I am with Nayan and Nyi. I know some of the answers to some of the questions I'll ask you below. You can explain what you do best though. So I'll ask anyways with the hope of providing a little grease on the communication wheels. > I'm Henrik Esbensen at DreamStart Labs > <http://www.dreamstartlabs.com/> . We are developing mobile solutions > to help people in developing countries realize their dreams of a > better life. Our first product is DreamSave > <http://www.dreamstartlabs.com/products.html> , a combination of > mobile and cloud technology which will digitize village savings > groups. DreamSave is being developed in partnership with Project Concern > International <https://www.pciglobal.org/> in Tanzania. It's wonderful to see a fintech focusing on savings. As important as loans are, I believe that the larger portion of missing banking coverage in developing countries is in the area of savings. What's mobile coverage like in Tanzania? How important are offline-first solutions there? Will you be working with smart phones (and which kind)? Will some of your services be provided via feature phones? Will you be using Fineract just for the core banking components, or are you supporting field officers as well? Are you using Fineract just to manage savings, customers, and groups, or are you also integrating with or implementing a mobile money system? [Henrik]: Mobile network coverage in Tanzania seems to be quite good. Most people I've met have two or more SIM cards, so that if one carrier doesn't work at any given time, they'll use another. So it is my impression that as long as you use more than one carrier, you will have coverage almost anywhere. We'll be using both smartphones and feature phones. The ideas is that each savings group has one smart phone which they share. Our app on this phone will do all the book keeping, manage the meetings, etc. And then each group member can query data from their feature phones. As for our use of Fineract, we want to go with Gen 3, and we are still in learning mode with respect to its exact capabilities/offerings. But we are interested in multiple ways, one of which is API's to mobile money, banks, credit scoring, etc. We are interested in the web app as well, in addition to core functionality, of course. You mention cloud technology, so I assume your team is acquainted with cloud hosting. Are you be deploying in one of the big cloud providers such as Google, AWS, or Azure? Are you deploying directly to one of these environments or are you using some sort of abstraction technology such as Docker, Kubernetes, OpenStack, or CloudFoundry? [Henrik]: We are running on Google's Cloud with Docker and Kubernetes and we are very happy with that setup. What do the privacy regulations look like in Tanzania? Do you have any special needs in this area? Are there any special needs in and around security in the communities that you serve? [Henrik]: Great question, and a complicated topic. We are still working on this ... - so I don't yet have specific, useful answers to this, but it's obviously important and we are treating it as such. It just takes time to figure out the legal and regulatory requirements, in any country, I think. > We have followed the development of Mifos Gen 3 (Mifos I/O) very > closely, and are very excited about it. Our team has attended many > Mifos meetings and webinars and had countless discussions with various > members of the Mifos community. It is clear that the next generation > architecture which Mifos I/O offers present very promising opportunities. As I've been answering these e-mails, this naming has been bothering me a bit. I find it potentially confusing. This software, while largely developed under the Mifos Initiative will, hopefully, see its first release under the auspices of Apache Fineract. Under Fineract this is not a "Gen 3"; we've only just brought out Apache Fineract 1.0.0. I can understand why the Mifos Initiative would call this Gen 3, but I am no longer associated with the Mifos Initiative. To avoid confusion, I'm going to skip numbering altogether, and call this Apache Fineract Cloud Native (or Fineract CN for short). If Mifos repackages this, Mifos can call it Mifos Gen3, but I will not be involved in those efforts. I'm glad to hear that you find our Fineract CN architecture convincing. What about our architecture fulfills needs that you have not been able to fulfill elsewhere? Is there anything in what you've already seen which could be changed to better fulfill your needs? [Henrik]: Thank you for clarifying the naming. Yes, I also find it a bit confusing. "Fineract CN" is good. We have just setup Fineract CN on Google Cloud now, and then we'll take a close look at the specifics. Once we have done that, I can give better answers to these questions. But for now, the more 'generic' answer is that Fineract CN has an industry-strength, scalable architecture. I believe that, at the present, you are dealing with a single tenant application; is that correct? Does this one tenant place enough scaling pressure on your application to justify the extra resources that a microservices architecture requires for a one instance install? Or are you hoping to attract more tenants? [Henrik]: Yes, we have a single tenant for now, but that will change. We expect to have more tenants in a foreseeable future. That's goes back to a key reason for going with Fineract CN: We expect to be needing the scalability. > We want to build technology solutions on top of the Gen 3 framework > and would love to provide input for the future roadmap. However, we > don't feel comfortable pursuing this path before the Gen 3 framework > is formally a part of Apache Fineract. Do I understand correctly that you wish to provide requirements, but not contribute code? [Henrik]: We are open to contribute code as well. Like any other startup, we don't have an abundance of resources 😊 But let's see which specific opportunities emerge once we get a bit further along. > So when will the framework be part of Apache Fineract? And how we can > be a part of this process? We are eager to see Mifos Gen 3 become a > part of the Apache Fineract project and would like to contribute as > individuals. > > > > We have worked extensively with the previous generation, Mifos X (Gen > 2), in the past. It was a great experience. At the same time, we are > well aware that Mifos I/O (Gen 3) will take the possibilities to an > even higher level, and we want to utilize that ASAP. The > micro-services architecture is very promising, as is various > integration possibilities. The opportunities for innovation on the Gen > 3 platform are very, very exciting. But it's difficult for me to see > how this can all come to full fruition before Gen 3 becomes a part of the > Apache Fineract project. I've been working for Kuelap, and not the Mifos Initiative since March of this year. The good news is that Kuelap is just as committed as Mifos is to contributing to Apache Fineract. All of my programming time is dedicated to the future Fineract CN code under https://github.com/mifosio. I have made nearly 500 commits in public to that source code since joining Kuelap. Many of those changes stem from our own experiences with cloud native deployment. If that code were under Apache, I'd be one of most prolific committers at Apache. Two other employees of Kuelap are providing even more intense dedication than I am. We are literally giving it everything we've got. The bad news is that that code is not quite far enough along to be ready for a first release. I wish to bring that code to first release status before we bring it into the Fineract community for two reasons: 1.) Right now we have tremendous pressure to achieve certain milestones which is common in startups. I do not have capacity for the various communication tasks required to bring this code into the community. 2.) I believe it will be easier to explain things and get people on board with code which has a first release level of maturity. We're close but we're not there yet. I hope you can be patient for a little longer. [Henrik]: Thanks for the info. This all makes sense. And I'm very impressed by the work you and Markus are doing. As mentioned above, we have just brought Fineract CN up on Google cloud in order for us to take a closer look. But maybe that's premature - should we wait a little? Thanks a lot, Myrle! Best Regards, Myrle Krantz
