Azure CosmosDB is too expensive for small but multiple collections
Currently using on-prem MongoDB (on Linux) and wish to move to Azure, but I find CosmosDB is too expensive for small but multiple (MongoDB)collections because it seems that a minimum of 400 RRU's/per second will be charged for each collection.
The terminology used on the pricing web pages is somewhat unclear though and I am not sure if the pricing for the minimum of 400 RRU's/second applies to partitions or collection (or if these terms are in fact identical semantically)
AzureEager
shared this idea
Cosmos DB supports sharing throughput across multiple collections with database-level throughput. This costs approximately $24/month.
Thank you for your feedback on this item. Will be closing this item at this time.
Thank you.
152 comments
-
Wick
commented
Merging all containers into one to avoid high consumption of RU/s sounds like a bad workaround - not to mention unique key constraints need to be dropped. As of now, I find your RU/s billing ridiculously expensive. I just looked at my metrics. A single user needs about 30 RU/s during the time they are logged in. How am I supposed to scale to millions of users? CosmosDB has become a showstopper, so I will be moving away from Azure soon.
-
E
commented
The official answer should be:
> Cosmos DB supports sharing throughput across multiple (max. 25) collections with database-level throughput.
In my case, I have 3 larger and 40 small collections (43 in total), the pricing would be at least 400 RU/s + 18 * 100 RU/s, which is 2200 RU/s and equals ~110EUR per month
https://docs.microsoft.com/en-us/azure/cosmos-db/set-throughput:
> In February 2020, we introduced a change that allows you to have a maximum of 25 containers in a shared throughput database, which better enables throughput sharing across the containers. After the first 25 containers, you can add more containers to the database only if they are provisioned with dedicated throughput, which is separate from the shared throughput of the database.
-
Rob Kraft
commented
Adding my voice to the chorus. RU pricing is too confusing. I understand that I can do 2 Selects and 3 Inserts per day to use up my 400 RU per month. That is ridiculous. I thought I misunderstood the pricing, until I got hit with the bill. Firebase and AWS DynamoDB are thousands of times cheaper.
-
Matt
commented
Microsoft, please reduce the ******** price on this trash. I'm paying $250 a month even though my consumption is probably a quarter of that. This is insane. Your pricing model is ripping off people.
-
Jan B
commented
"Cosmos DB supports sharing throughput across multiple collections with database-level throughput."
Can you please point us to the relevant documentation for this feature as apparently it's not obvious how to enable this (going by recent comments)?
-
Eric Tijerina
commented
Just got slapped with a $175 bill for 3 collections with less than 1MB !!! What the actual **** is this?!
How are developers able to run Mongo on azure without going out of business??
-
W
commented
Everyone here is lucky there's now MongoDB Atlas :)
-
Tommy Tucker
commented
@Roman why are you not combining those containers and adding a property if _type to each document? With that few of documents it shouldn't effect query speed. Query on _type when you need only those documents. Your database will shrink to 1 to 3 containers.
-
Roman
commented
CosmosDB is becoming prohibitively expensive. My application needs about 15 containers, of which 12 have less than 50 rows each. Yet, I am forced to pay for 100RU for each of those containers, which makes this project look really stupid... I am paying from my own pocket to run this for a charity volunteer center. Now guess what - I had to merge some of those containers together, mess up the code in the process - just to work around the greedy MS.
-
Dan Jarvis
commented
I agree that converting an application to use sharded collections is a bit much to take advantage of the per-database throughput plan, though I can understand why technically it needs to be that way. For our application it'd be quite an effort, taking into consideration all the QA regression and performance testing and so forth.
-
Anonymous
commented
I had the same exact question and I don't feel the answer could have been clearer. I would also like to mention that there are lots of applications with small datasets that could take advantage of a more generous pricing scheme. How about free until the database gets to be a certain size?
-
Szymon Sasin
commented
Any tutorial how to enable it?
-
Anonymous
commented
I vaguely rembered coming across this issue - after I came across this issue again.
The default pricing is plain horrible - I just spun up a trial instance and ran my app against it (instead of Mongo DB Atlas, which is free). Starting my app created a few collections (1-2 records in each) - those few kilobytes already would have put me in the 3-digit range when it comes to pricing. Seriously, guys...
-> Provide a free tier for DEV / Eval purposes
-> Make it easier to setup per-database pricing with much lower minimum RU/s. I don't know why that would be a problem - don't worry, if we need additional RU/s, we will buy them, obviously... -
Anonymous
commented
I totally agree with John's comment
________________________________________________I totally agree with the original comment. We too were burned with a tiny Cosmos system we were using to prototype before scaling up. We have completely turned it off.
The pricing is very unclear. You can set up a shared database but then have to code to use shards across different collections. The RU/s measure is really confusing. You are given a free 100 RU/s tier when you sign up, but the minimum is 400 RU/s, so the "free" addition is meaningless.
We originally had 8 collections with the intent to grow to about 20. By default it put 1000 RU/s into each one (we didn't understand the complex pricing). The cost per day for 1000 RU/s is $1.92. 20 collections x 30 days x $1.92 = $1,152 per month for a database that was about 100 MB. The equivalent on Atlas MongoDB is zero.
I raised the issue on Microsoft support and the initial answer was that they didn't know about Cosmos.
I cannot understand why anyone would use this. My recommendation is go elsewhere.
____________________________________________________________
The issue is not really resolved. I don't want to code with shard, it's complex. Please give a real solution.
-
ShanHuiyang
commented
Cosmos DB supports sharing throughput across multiple collections with database-level throughput. This costs approximately $24/month.
___________________________________________________________________________________________________
The problem is, how to do that?? I did not find any settings on Azure portal to do that.
-
Jeff Johnson
commented
This is not my experience. It seems to reserve 400 RRU, which is about 125 USD / month. Why can you not start with a dynamic amount of RRU, why do you have to reserve. If you only use 10 RRU in a month you should pay for 10, not 400. Very strange pricing...
-
Joel Oughton
commented
If you're using .NET you can use Nebula to reduce the cost of Cosmos DB. All documents are stored within a single collection with logical separation. This avoids the cost racking up when lots of different document types need storing. You also get other features such as a simplified versioning pattern.
https://github.com/cloud-maker-ai/Nebula
Disclaimer: I'm the author.
-
John
commented
Why did you close this feature request? Clearly, Microsoft does not care about its customers. Please reduce the cost of provisioning 100 RUs per second. Any serious app needs multiple containers (at least 20). We are being charged for containers whether we use them or not. The current pricing model sucks.
-
Zachary
commented
Really? $24/month is too expensive for my church project. I was hoping to use Azure since we don't have the money for our own hardware. I thought the point of the cloud was to support projects at any scale. This is very discouraging.
-
Anonymous
commented
Currently we r facing lots of RU limit isue with cosmosdb. For a 5 gb of small db i had to make it run as 10.000 RU and this is too expensive. Also monitoring of cosmos is wrong or Cosmos has some bug. Although monitor shows that i am very under the limits, i am getting 429 which is rate limit error. I realy couldn't understand how it is a big data solution. Should i pay tens of 1K USD? By the way we are on Gremlin API