Listen to this Post
Skip to search
Skip to main content
Keyboard shortcuts
Close jump menu
Search
Tony Moukbel
Feed detail update
Feed post
View Yan Cui’s graphic link
Yan CuiYan Cui
• 3rd+3rd+
I help teams build Production-Ready Serverless systems | AWS Serverless HeroI help teams build Production-Ready Serverless systems | AWS Serverless Hero
View my newsletter
4h • 4 hours ago • Visible to anyone on or off LinkedIn
Follow
Mm.. Aurora DSQL is surprisingly slow for complex joints. 8-9s vs. 4-5ms for Aurora
Thanks László Bodor for sharing!
Activate to view larger image,
No alternative text description for this image
Activate to view larger image,
likeinsightfullove
67
4 comments
Reactions
View Kurt Lee’s graphic link
like
View Jakub Barczyk’s graphic link
like
View Aurélien Perrier ☁️’s graphic link
like
View Arif Akkermans’ graphic link
like
View Pedro Barbosa’s graphic link
insightful
+62
Photo of Tony Moukbel
Thanks for sharing, Yan
Thoughtful post, thanks Yan
Well put, Yan
Insightful
Definitely worth reading
Love this, Yan
Great breakdown
Add a comment…
Open Emoji Keyboard
Current selected sort order is Most relevant
Most relevant
View Yevhenii Dytyniuk’s graphic link
Yevhenii Dytyniuk
• 3rd+
AWS Certified Architect. Kubernetes. Serverless. Event-Driven and Service-Oriented Architectures
now
nerd_mode=On
Can I see the EXPLAIN – query execution plan?
From the query itself, I do not see the need to LEFT JOIN for product categories. LEFT JOIN would always make DB engine to run a full scan on the main table. Is there a chance to have a product without a category? Also, are products never deleted? Can’t the order_items table be denormalized and hold product data within the order item? So, order lines retrieval is a matter of querying a single table? The same for user shipping address – what if a user deletes their shippind address? The order data is incomplete. It could be wise to bake immutable order information into respective rows.
Questions, so many questions… 😅
I’m speculating, but given a meaning of the D in DSQL, I presume, that individual tables are stored in different partitions, thus cross-partition JOIN would be expensive.
…more
Like
like
1
Reply
1 reply
1 Comment on Yevhenii Dytyniuk’s comment
View László Bodor’s graphic link
László Bodor
• 3rd+
Cloud Architect | 6x AWS Certified | First and foremost a builder!
now
Yevhenii Dytyniuk sure the query can be optimized. The point was to illustrate the difference in speed compared to Aurora Standard where the same query ran in ms range.
Like
Reply
View Andrew W.’s graphic link
Andrew W.
• 3rd+
Full Stack Developer / Technical Lead @ Take Command
now
I’d be interested in seeing a comparison between Aurora DSQL vs something like Neon
Like
like
2
Reply
View Vladimir Samoylov’s graphic link
Vladimir Samoylov
• 2nd
Predictable & Unlockable Problem Solver | Cloud Architect | Co-Owner @FivexL | Founder @ZoJump | AWS Community Builder | DevSecAI | AI/ML | RAG/LLM
now
Thanks for sharing
I think that more DSQL real world tests and example would help
Like
like
1
Reply
References:
Reported By: Theburningmonk Mm – Hackers Feeds
Extra Hub: Undercode MoN
Basic Verification: Pass ✅
