[Ongoing series of blog posts
to inform potential developers, users and (hopefully investors) about this new app ecosystem I am architecting, designing, developing and deploying. More details at this page]
I have learnt (perhaps the hard way) that a project without a deadline, is not a project at all! Sure, the deadline may get extended, and the project may die away. That’s the negativity in me speaking. Of course, the deadline could be met, and the project may actually reach something that resembles completion.
Project TD needs to hit beta phase
in less than 10 months. The official launch is less than 6 months after that. I am racing towards a deadline, and that means, I need to pick up a starting point. The requirements planning is all good and fine, but at the end of day, some programming needs to be done. The more I think about it, the more I feel that the API Engine is where it all starts.
That brings up the question - what is the API Engine? Everything and anything that works with some service is an API. For instance, the app ecosystem will use the Facebook provided login system on the end user side. The admin folks will use a Microsoft Identity library. I do not think I will bother with providing alternate login systems for end users. This does limit access to some extent. For instance, users who don’t have a facebook account might choose not to be part of Project TD.
That should be okay. As I have already mentioned
, this entire project will be open source non-commercial usage. So, if someone wants to include some other login system, the design is modular enough to allow another developer to latch on Google or Twitter or email registration based system. I must understand my limitations here. For the foreseeable future, this entire ecosystem will be built by me. If there is an opportunity that can be exploited and work load can be reduced. The login system, at least for the end user, I am going to limit myself to Facebook. I like Facebook. The Graph API is simple to use, works on almost any platform and it has the color blue.
That’s the story with the login system. And the above Facebook login system is made possible because of the extensive login API’s that the Graph API
(Facebook branding for their API stuff) provides.
Anyway coming back, the API Engine will be the collection of all the API calls, the entire ecosystem will make. That includes the many apps that will be used by the restaurant patrons and the owners, the admins, the ops folks and other roles that might be included as the design evolves. The API Engine is the only way the users can interact with the database, which stores the data. Now, the API engine (which is essentially server that listens and responds to calls from the users) is the store front for all the apps. On the other side, the API Engine interacts with the database, processing all the stuff that the calls are asking for.
This allows for a breakdown of responsibilities, which is essentially, the whole point of modularity. Thanks to this, it is possible for me to design redundancy around the database. Build backups, replace when data goes corrupt (or gets hacked) and replace, hopefully on the fly. Since everything is broken up into pieces and connected with bridges. The source and destination don’t really care where the bridges actually go on the other side. All they care is that, on their side, the bridge is fine.
This API Engine, is like an island at the very center. A huge island that has everything. Below the island (the API Engine) is the underwater volcano. Think of this as the database. Now, this huge island is surrounded by small water things like tiny islands, ships, ports and boats. All of these folks need something but they cannot talk to each other directly (that is whole point because boats die over time. We don't want the bridges to die with them, bringing the entire system down with them). They are all connected to the island, and everything goes through it. The island, on its own is nothing without the underwater volcano (the database).
Thinking backwards, the center of this whole ecosystem would be the island and the volcano under it. Once these two are in place, the rest – boats, ships, sea gulls, ports, sharks and dolphins – will start settling around it. Everything in the system can be replaced of course. The island, volcano, the hundreds of boats, ships and sharks. Redundancy is the name of the game man!
As a consequence, I am now spending all my time designing the list of APIs that will from the first set of API calls, which together will become the API engine. As with the island-volcano thing from above, the API engine is useless without a database. So, along with the API engine, I will also have to design the data storage system. For both the database and the server that will respond to API calls, I am going with Microsoft Azure services. I have enough experience dealing with Microsoft Azure, and it has every possible thing imaginable to build and deploy pretty much every aspect of the ecosystem. There are alternatives (Amazon comes to my mind) but I simply have had better exposure to Microsoft services, and despite their higher charges, I will go with that.
As I design (and then build the API Engine and the database), I figure that I will need an API dashboard that will keep track of the API calls currently being designed, developed, tested, deployed and available. This should become extremely useful to allow the stakeholders to keep track of what is happening, and obviously, what is not happening.
The dashboard will be the choice of tool for the operations team whose job is to make sure that the system is working as designed. This also means, they need to be provided with the necessary information to deal with problems when they occur. Then, when the issues are fixed, the data should reflect the resolution. This also, will be routed through the API calls, and will be part of the API engine. The smart thing to do would be to separate the API Engine to multiple parts. However, if the entire system is backed up (and it is simple enough) then it makes sense to maintain just one engine, and then have backup engines ready to go when things go bad.
The interesting thing here is that, the island and volcanos are separate. Each of them have copies of them, waiting in the pipeline if disaster should strike. If I split the engine into multiple parts, the redundancy maintenance factor will go up by that much. For now, it seems wiser to simply have one system to maintain, instead of multiple specialised systems.
Next up, I must pick up the design tools which is where my new iPad and the pre-owned printer comes into the picture. I will talk about that in the next blog post.