Google offers the Firebase suite of application development tools, available free or at cost for higher degree of usages, priced flexibly accorded to features needed. The suite includes A/B testing and Crashlytics, Cloud Messaging (FCM) and in-app messaging, cloud storage and NoSQL storage (Cloud Firestore and Firestore Realtime Database), and other features supporting developers with flexible mobile application development.
Well suited: To most of the local run of datasets and non-prod systems - scalability is not a problem at all. Including data from multiple types of data sources is an added advantage. MLlib is a decently nice built-in library that can be used for most of the ML tasks. Less appropriate: We had to work on a RecSys where the music dataset that we used was around 300+Gb in size. We faced memory-based issues. Few times we also got memory errors. Also the MLlib library does not have support for advanced analytics and deep-learning frameworks support. Understanding the internals of the working of Apache Spark for beginners is highly not possible.
Firebase should be your first choice if your platform is mobile first. Firebase's mobile platform support for client-side applications is second to none, and I cannot think of a comparable cross-platform toolkit. Firebase also integrates well with your server-side solution, meaning that you can plug Firebase into your existing app architecture with minimal effort.
Firebase lags behind on the desktop, however. Although macOS support is rapidly catching up, full Windows support is a glaring omission for most Firebase features. This means that if your platform targets Windows, you will need to implement the client functionality manually using Firebase's web APIs and wrappers, or look for another solution.
Analytics wise, retention is extremely important to our app, therefore we take advantage of the cohort analysis to see the impact of our middle funnel (retargeting, push, email) efforts affect the percent of users that come back into the app. Firebase allows us to easily segment these this data and look at a running average based on certain dates.
When it comes to any mobile app, a deep linking strategy is essential to any apps success. With Firebase's Dynamic Links, we are able to share dynamic links (recognize user device) that are able to redirect to in-app content. These deep links allow users to share other deep-linked content with friends, that also have link preview assets.
Firebase allows users to effectively track events, funnels, and MAUs. With this simple event tracking feature, users can put organize these events into funnels of their main user flows (e.g., checkout flows, onboarding flows, etc.), and subsequently be able to understand where the drop-off is in the funnel and then prioritize areas of the funnel to fix. Also, MAU is important to be able to tell if you are bringing in new users and what's the active volume for each platform (Android, iOS).
Attribution and specifically multi-touch attribution could be more robust such as Branch or Appsflyer but understand this isn't Firebases bread and butter.
More parameters. Firebase allows you to track tons of events (believe it's up to 50 or so) but the parameters of the events it only allows you to track 5 which is so messily and unbelievable. So you're able to get good high-level data but if you want to get granular with the events and actions are taken on your app to get real data insight you either have to go with a paid data analytics platform or bring on someone that's an expert in SQL to go through Big Query.
City-specific data instead of just country-specific data would have been a huge plus as well.
If the team looking to use Apache Spark is not used to debug and tweak settings for jobs to ensure maximum optimizations, it can be frustrating. However, the documentation and the support of the community on the internet can help resolve most issues. Moreover, it is highly configurable and it integrates with different tools (eg: it can be used by dbt core), which increase the scenarios where it can be used
Firebase functions are more difficult to use, there are no concepts of triggers or cascading deletes without the use of Firebase functions. Firebase functions can run forever if not written correctly and cause billing nightmares. While this hasn't happened to us specifically it is a thing that happens more than one realizes.
1. It integrates very well with scala or python. 2. It's very easy to understand SQL interoperability. 3. Apache is way faster than the other competitive technologies. 4. The support from the Apache community is very huge for Spark. 5. Execution times are faster as compared to others. 6. There are a large number of forums available for Apache Spark. 7. The code availability for Apache Spark is simpler and easy to gain access to. 8. Many organizations use Apache Spark, so many solutions are available for existing applications.
Our analytics folks handled the majority of the communication when it came to customer service, but as far as I was aware, the support we got was pretty good. When we had an issue, we were able to reach out and get support in a timely fashion. Firebase was easy to reach and reasonably available to assist when needed.
Spark in comparison to similar technologies ends up being a one stop shop. You can achieve so much with this one framework instead of having to stitch and weave multiple technologies from the Hadoop stack, all while getting incredibility performance, minimal boilerplate, and getting the ability to write your application in the language of your choosing.
Before using Firebase, we exclusively used self hosted database services. Using Firebase has allowed us to reduce reliance on single points of failure and systems that are difficult to scale. Additionally, Firebase is much easier to set up and use than any sort of self hosted database. This simplicity has allowed us to try features that we might not have based on the amount of work they required in the past.
Makes building real-time interfaces easy to do at scale with no backend involvement.
Very low pricing for small companies and green-fields projects.
Lack of support for more complicated queries needs to be managed by users and often forces strange architecture choices for data to enable it to be easily accessed.