Ayush Pandey

Ayush Pandey

Distributed systems researcher

Fallacies of distributed computing

less than 1 minute read

Published:

Today while looking at some literature about serverless computing from RedHat, i came across a wikipedia article that mentions the fallacies of distributed computing. The original assertions were made by L Peter Deustch and others at Sun Microsystems.

Having worked with distributed systems for a little while now, I think i have made all of these mistakes when thinking about a system.

The fallacies are the following:

  1. The network is reliable
  2. Latency is zero
  3. Bandwidth is infinite
  4. The network is secure
  5. Topology does not change
  6. There is one administrator
  7. Transport cost is zero
  8. The network is homogenous

Throughout the years, there have been additions to this list. Mostly as system design paradigms have evolved. With serverless computing, there is three additional fallacies which are relevant.

  1. Versioning is simple
  2. Compensating transactions always work
  3. Observability is optional

I am now exploring this domain of computing for future research challenges. More details soon.

You May Also Enjoy

Consistency models and their interplay

11 minute read

Published:

Consistency models are difficult to explain and even more difficult to understand. At the time of writing, there is about 20 consistency levels that I know of, each with its own set of assumptions and implications which in a lot of cases are only subtly different between different consistency levels. Adding to this complexity is the fact that different communities refer to different things with the same name. Take Snapshot Isolation(SI) for example. To a relational database practitioner, SI means concurrent snapshots that remain isolated from each other. It is not really a consistency level but rather the Isolation property in the ACID mode. But for a distributed systems practitioner, SI means something completely different.

The World of Graph Databases from An Industry Perspective

5 minute read

Published:

This paper by Yuanyuan Tian presents an overview of the graph database landscape from an industry perspective. It covers the graph database landscape, graph query languages and several graph database products. It also discusses the challenges and opportunities in the graph database industry.

Dominance Locking in arbitrary rooted hierarchies

4 minute read

Published:

The current state of affairs in databases is typically oriented towards two-dimensional data. If some data is managed in an apparent multidimensional space, then it is usually achieved by some trick that involves joining tables and pivoting them. This not only makes performing queries on the data difficult and less efficient but also introduces an overhead of managing these multidimensional views.

DomLock: a new multi-granularity locking technique for hierarchies

7 minute read

Published:

In the paper “DomLock: a new multi-granularity locking technique for hierarchies” the authors present an interesting technique for locking objects that follow a hierarchical structure. Several applications that operate on objects that form a hierarchy suffer from synchronization bottlenecks because of lock contention. When several transactions operate on different parts of the data hierarchy, there is contention which restricts concurrency.