Excellent overview of Timestream, and love seeing Kotlin! One minor correction: Timestream has a configuration to allow direct write to magnetic store. You state that writes are prohibited outside the memory range; the docs say otherwise. Thanks again, this was really helpful.
Thanks Wolfgang! I didn't delve into performance benchmarks for this video unfortunately, that's something I'm curious about as well. I'll be someone has done a comparison!
Is timestream good to be read directly from a web api for aggregate timeseries data? I was wondering if I should use timestream aggregate data directly or send timestream aggregated data to dynamodb to be read from the front end.
Generally speaking yes I think you can use it to hydrate user facing pages, as long as it is sufficient in terms of the performance you require for your application. Re: aggregations I'd probably first try to use Timestream directly and see if it suits your needs from a cost and performance perspective. Scalability shouldn't be an issue. But I'm not sure how it compares to DDB in terms of cost and performance for potentially high volume user facing reads. One thing I'd definitely avoid is storing the raw data in DDB and performing aggregations on data retrieved from DDB.
Thanks and thanks for watching! The Timestream pricing page throws a bit of a curveball - it lists the magnetic storage price per *month*, but the in-memory storage price per *hour*. So to facilitate a more apples to apples comparison, I multiplied the $0.036/hour price by 720 (the hours in a month) and presented it that way.
Hey Rojan - it's been awhile since I've used Timestream, I believe you can store JSON data, but I'm pretty sure you can't efficiently query based on the contents of that JSON data. I'm not 100% sure about that though.
Very cool video. I used Amazon Timestream in the past, but I've been having lots of fun lately with QuestDB, an open source time series database. Do you have any views on it? Thanks!
![AWS re:Invent 2021 - Deep dive on Amazon Timestream [REPEAT]](/img/1.gif)
I admire the thought and time put into these videos. Well done.
thank you! this was my first one ever.
Excellent overview of Timestream, and love seeing Kotlin! One minor correction: Timestream has a configuration to allow direct write to magnetic store. You state that writes are prohibited outside the memory range; the docs say otherwise. Thanks again, this was really helpful.
Thanks for pointing this out! I wonder if that's something that's changed since I made the video...
Great overview of the pros and cons of different ways of storing and analyzing time series data. Super helpful!
Glad you liked it!
Thanks for giving the exact overview I was looking for. Will try it out.
nice, glad you found it valuable! this was the first video I ever posted, really cool to see that it's still being watched 😎
Comparison of different data storages is very useful.
glad you found it valuable!
As a beginner this has been super helpful and clarified a lot of issues for. Thanks !!
wonderful explanation, your storytelling approach is very good. please make a lot of vidoes.
Thank you, glad you found it valuable! Making more videos is the plan. What sort of topics would you like to see covered?
good overview, i'm missing a bit performance difference because of optimized indexes, and also time based aggregates
Thanks Wolfgang! I didn't delve into performance benchmarks for this video unfortunately, that's something I'm curious about as well. I'll be someone has done a comparison!
Excellent video
thanks, glad you found it valuable! this was the first one I ever posted - it's fun to look back on!
There is a minimal cost to the query which can also balloon if you need frequent queries.
great point!
Is timestream good to be read directly from a web api for aggregate timeseries data? I was wondering if I should use timestream aggregate data directly or send timestream aggregated data to dynamodb to be read from the front end.
Generally speaking yes I think you can use it to hydrate user facing pages, as long as it is sufficient in terms of the performance you require for your application. Re: aggregations I'd probably first try to use Timestream directly and see if it suits your needs from a cost and performance perspective. Scalability shouldn't be an issue. But I'm not sure how it compares to DDB in terms of cost and performance for potentially high volume user facing reads. One thing I'd definitely avoid is storing the raw data in DDB and performing aggregations on data retrieved from DDB.
Great video!
thanks, glad you found it valuable!
Great Video !
> at 8:43 there's a mistake in the image, it says $26.28 instead of $0.036
Thanks and thanks for watching!
The Timestream pricing page throws a bit of a curveball - it lists the magnetic storage price per *month*, but the in-memory storage price per *hour*. So to facilitate a more apples to apples comparison, I multiplied the $0.036/hour price by 720 (the hours in a month) and presented it that way.
Dynamodb with multiple primary keys has been doing great for me (no aggregations at all)
Nice! yeah not needing aggregations really opens up the possibilities...
AWS Timestream allows JSON structure data ? is it possible to fetch the query by applying condition on saved JSON data?
Hey Rojan - it's been awhile since I've used Timestream, I believe you can store JSON data, but I'm pretty sure you can't efficiently query based on the contents of that JSON data. I'm not 100% sure about that though.
Athena serverless is best for querying json
Nice video..Short and very helpful!
thanks glad you found it valuable!
Very nice analysis! Thank you.
Glad you found it valuable, thanks for watching!
Very cool video. I used Amazon Timestream in the past, but I've been having lots of fun lately with QuestDB, an open source time series database. Do you have any views on it? Thanks!
how did you host QuestDB on AWS,? probably on a ec2 which brings extra cost