1. Lack of support for geo-distributed applications and inability to maintain application availability during planned events are two major barriers preventing sharding frameworks from being adopted by the majority of sharded applications.
2. Supporting many complex applications in a one-size-fit-all sharding framework and the difficulty in supporting sophisticated shard-placement requirements are other adoption barriers.
3. Facebook's Shard Manager has overcome these adoption barriers, currently being used by hundreds of applications running on over one million machines, accounting for 54% of all sharded applications at Facebook.
The article is generally reliable and trustworthy, as it provides evidence to back up its claims and presents both sides of the argument fairly. The authors provide references to relevant research papers and studies that support their claims, which adds credibility to their arguments. Additionally, the authors provide an overview of the challenges faced by existing sharding frameworks and how Facebook's Shard Manager has addressed them.
However, there are some potential biases in the article that should be noted. For example, the authors focus mainly on Facebook's Shard Manager and do not discuss other existing solutions or frameworks that could potentially address the same issues. Additionally, while they mention other adoption barriers such as supporting many complex applications in a one-size-fit-all sharding framework and difficulty in supporting sophisticated shard placement requirements, they do not provide any evidence or examples to illustrate these points further. Furthermore, while they mention that a constraint solver can theoretically handle complex placement requirements, they do not explore this option further or discuss any potential drawbacks or limitations associated with it.
In conclusion, while this article is generally reliable and trustworthy due to its use of evidence to back up its claims and fair presentation of both sides of the argument, there are some potential biases that should be noted when evaluating its trustworthiness and reliability.