Sitemap

Software Engineering Dose #2

4 min readDec 11, 2022

--

Hey All, Hope you like the first edition of this digest. For those who missed

In this edition, I am sharing the new set of articles which I read over the past one week. Let’s get into it…

Managing PostgreSQL indexes

In the first one, Knock labs shares some of the findings they rely on while managing Postgres Indexes.

Did you know that if you have an index on the column, Postgres might still choose to ignore the index and do the heap scan for an indexed column? It all comes down to the query planner determining the best way to fetch data.

They also talk about different types of indexed and what each of them signifies and also combined indexes vs multi-column indexes. Really interesting read.

Serving Personalised offers at Scale

Games24x7 Engineering shares a detailed article on serving offers to customers at scale.

They shared the full package actually like the feature requirements, tradeoffs, high-level architecture, the issues they faced while testing and the actions taken, etc. Also, I feel this is something that is not just specific to Games24x7 Engineering but we can apply this to a wider range.

For example, have you wondered how some of the offers on different websites like Swiggy, Zomato, Amazon, etc are specific to a few individuals? Well, here is how they do it.

Press enter or click to view image in full size
High level architecture

Scaling Customer review system at Booking.com

In this article, we can see the production use of consistent hashing for storing data at scale.

Get Sriteja Parimi’s stories in your inbox

Join Medium for free to get updates from this writer.

For better latency, they created a materialized view of Reviews data and used shards to store that data. Basically like a key value structure, where all of the reviews data is stored. The challenge comes when they have to increase or decrease the number of shards on demand.

Using a normal modulo hashing would require them to move almost all of the data for every upscale or downscale. They used a different type of consistent hashing algo called Jump consistent hashing for this case. The main advantage being the minimal amount of data being reassigned/moved for the up or down scale.

If you want to read up on Jump consistent hashing, here is the link. Jump consistent hashing is a really fascinating topic.

Be careful before you turn on HTTP/2

Interesting article from Lucidchart tech on why turning on HTTP/2 was a mistake. We can add this one to the Things to keep in mind before turning on HTTP/2.

Real time analytics using Druid by Netflix

Netflix Technology Blog uses Druid to deliver Real time analytics.

They share their whole architecture from client to storage where they talk about each component in detail. One direct application of these insights is to create alarm signals on these events to alert the responsible team for any possible issues.

They also talk about how they ingest the data and how they handle live and historical data and the compaction process. This can also be applied to Top k searches in the last X minutes kind of system design questions.

Revamping Search architecture at Canva

Canva Engineering explains the challenges faced in existing Search architecture and how they migrated to new architecture, the decisions behind the new one, the tradeoffs, etc.

This is shared in 2 parts where in the first part they share the problems and the requirements for the new system and the second one has the architecture and explanation of various components in the architecture.

Press enter or click to view image in full size
Canva revamped Search architecture

That’s it for this time folks. I really enjoyed reading these articles and preparing notes and hope you find these articles knowledgeable.

Please also share any articles from your Interesting read/watch list in the comments. Let’s upskill together!!!

--

--

Sriteja Parimi
Sriteja Parimi

Written by Sriteja Parimi

Software Developer. Writes about software engineering. Constant learner.