I am never sure I get the unit conversions correct. I've seen folks say it costs about $4/month to r

I am never sure I get the unit conversions correct. I've seen folks say it costs about $4/month to run a DO 24/7, but I've also seen $0.34/day which works out to about $10/month. Can someone show me the math?

LLarry I am never sure I get the unit conversions correct. I've seen folks say it costs...

lambrospetrou•10/27/25, 6:37 PM

1 DO running 24/7 needs:

0.128 GB-s * 3600 seconds_per_hour * 744 hours_in_month = 342_835 GB-s/month

0.128 GB-s * 3600 seconds_per_hour * 744 hours_in_month = 342_835 GB-s/month

and at the price

$12.50/million GB-s

$12.50/million GB-s

that's 4.285 USD/month (

342_835/1_000_000*12.50

342_835/1_000_000*12.50

).
You get 400,000 GB-s for free.

The above is duration only charges.

edit: using the correct

0.128 GB-s

0.128 GB-s

0.125GB-s

0.125GB-s

LarryOP•10/27/25, 6:38 PM

Thanks! I only need duration charge help. I can count requests, just not unit convert.

PhoenixSpace•10/27/25, 6:53 PM

Is there any way to get a hook to run code before a DO is evicted from memory? Would love to use it to flush in memory stuff to sql storage rather than needing to write a value that will be modified often in short bursts.

PPhoenixSpace Is there any way to get a hook to run code before a DO is evicted from memory? W...

LarryOP•10/27/25, 6:56 PM

There is no such hook and even if there were, it wouldn't always work because it can evicted for crazy reasons. The saving grace is that reading from and writing to storage is very fast due to caching and locality. I have switched to a coding style where I don't have instance variables unless the in-memory representation is an expensive transformation from the in-storage representation. It's counter-intuitive but think of storage like you used to think of instance variables.

LarryOP•10/27/25, 7:04 PM

For the cases where the in-memory transformation is expensive, I have also adopted a coding style where I save and retrieve that largish structure in one whole shot using the fact that the KV API let's you store just about everything (objects with cycles or aliases, Date, Map, Set, Infinity, NaN, etc.). See the complete list in the second column here

LLarry There is no such hook and even if there were, it wouldn't always work because it...

PhoenixSpace•10/27/25, 7:05 PM

That's pretty fair, wishful thinking I guess. Wanted to avoid the $1/million write charges because it's an extremely bursty workflow (counting a rapid stream of events), but honestly not that bad

PhoenixSpace•10/27/25, 7:05 PM

the storage backend KV-like API for sqlite is OP though

LarryOP•10/27/25, 7:08 PM

Reads are so inexpensive I ignore the cost. Write costs can add up.

PhoenixSpace•10/27/25, 7:08 PM

Yup, that's why I was wondering if I could just flush it with a handle

PhoenixSpace•10/27/25, 7:09 PM

Guess this might be a better task for a container

LarryOP•10/27/25, 7:10 PM

I don't know the costs for containers but I find it hard to image it would be less under the vast majority of circumstances.

PhoenixSpace•10/27/25, 7:56 PM

On average it will undergo ~250,000 events in a 8 minute span, and then go silent for up to a month, so about $0.25 in writes vs probably a couple cents for the container

PhoenixSpace•10/27/25, 7:56 PM

very niche/weird usecase lol

PPhoenixSpace On average it will undergo ~250,000 events in a 8 minute span, and then go silen...

1984 Ford Laser•10/27/25, 8:24 PM

Could the events be grouped in some way? Store groups of events per write request

LarryOP•10/27/25, 8:27 PM

For $0.25, I wouldn't bother optimizing. Not even for $25. Maybe at $250 if it was simple to implement and manage but containers are generally harder to manage in my experience.

11984 Ford Laser Could the events be grouped in some way? Store groups of events per write reques...

PhoenixSpace•10/27/25, 8:39 PM

Unfortunately I don’t control that side. Been asking for that, but they don’t want to do it

LLarry For $0.25, I wouldn't bother optimizing. Not even for $25. Maybe at $250 if it w...

PhoenixSpace•10/27/25, 8:41 PM

I totally agree, but I’m basically donating the cost to manage this out of pocket currently. I was originally going to just spin up a VPS but figured scale to zero could save a bit (and scale horizontally since I don’t know how many events per month will be happening initially)

PhoenixSpace•10/27/25, 8:42 PM

It’s for managing some stuff for a highschool competition being spun up. Infra is a bit of a mess but until it goes public and gets sponsored basically just doing it for free.

PPhoenixSpace It’s for managing some stuff for a highschool competition being spun up. Infra i...

1984 Ford Laser•10/27/25, 8:44 PM

If it is actually only once a month then you'll stay within the included limits with workers paid anyway

1984 Ford Laser•10/27/25, 8:45 PM

Doing lots of writes sucks but I wouldnt worry unless your use case would easily push you over included limits

11984 Ford Laser If it is actually only once a month then you'll stay within the included limits ...

PhoenixSpace•10/27/25, 8:45 PM

It’s once a month per team per event

1984 Ford Laser•10/27/25, 8:45 PM

ahhh

PhoenixSpace•10/27/25, 8:45 PM

And I have no clue what the number of teams or events will be

1984 Ford Laser•10/27/25, 8:45 PM

THat might be an issue lol

PhoenixSpace•10/27/25, 8:46 PM

Yeah

1984 Ford Laser•10/27/25, 8:46 PM

What are you doing with the data afterwards?

PhoenixSpace•10/27/25, 8:46 PM

It gets processed into a leaderboard, basically

PhoenixSpace•10/27/25, 8:46 PM

Hmm. It just occurred to me this could totally be done with WAE

PhoenixSpace•10/27/25, 8:47 PM

Probably much simpler

1984 Ford Laser•10/27/25, 8:47 PM

I wonder if you could store the entire 300k events as a single object in R2 then just process them when you need them via streaming into a Worker?

PPhoenixSpace Hmm. It just occurred to me this could totally be done with WAE

1984 Ford Laser•10/27/25, 8:47 PM

or this maybe

PhoenixSpace•10/27/25, 8:48 PM

Yeah, I’ll probably just use WAE lmao

1984 Ford Laser•10/27/25, 8:48 PM

300k events for a leaderboard is wild to me, what are you doing exactly lol? If youre allowed to talk a bit more about it

1984 Ford Laser•10/27/25, 8:49 PM

ahh I see WAE is only 0.25 per mil writes

PhoenixSpace•10/27/25, 8:50 PM

Honestly I don’t know a lot of details. Friend came to me and asked for help with the data pipeline. As far as I know it’s some tech related challenge thing, and certain events generate milestones they have to track with some light validation in worker. AFAIK their reasoning for not batching was concern someone would be able to edit it on the computer before the batch is pushed (this seems like something you can design around but I’m not making that part)

11984 Ford Laser ahh I see WAE is only 0.25 per mil writes

1984 Ford Laser•10/27/25, 8:52 PM

oh and you only get charged per read query not per row thats kinda crazy

PhoenixSpace•10/27/25, 8:53 PM

tbh it sounded like a fun excuse to do more with cloudflare and that’s all I really paid attention to

PPhoenixSpace tbh it sounded like a fun excuse to do more with cloudflare and that’s all I rea...

1984 Ford Laser•10/27/25, 8:53 PM

Honestly if you ignored costs there's probably a dozen different ways to use CF products to do this

1984 Ford Laser•10/27/25, 8:53 PM

I'll always take more opportunities to build diff stuff on CF

PhoenixSpace•10/27/25, 8:54 PM

Same

PPhoenixSpace Yeah, I’ll probably just use WAE lmao

Hard@Work•10/27/25, 9:01 PM

Though note WAE can perform sampling

LLarry I had a chance over the last week to do a deep dive on Cap'n Web. Here is my mul...

1984 Ford Laser•10/27/25, 9:23 PM

I found this really interesting, not only to help me understand more generally how RPC works under the hood, but also how I can improve some of my dx and implement useful patterns like OCAN. Definitely can't wait to see where you take Lumenize, unless CapnWeb takes over (which it seems like you aren't convinced on the standard yet)

hollowaya•10/27/25, 9:56 PM

does anyone know if the KV backend DOs have a 30s limit, or if they also respect the 5m limit if set? docs are kinda ambiguous

hollowaya•10/27/25, 10:56 PM

also, does anyone know if memory or logging of out-of-memory are improved on the sqlite backend?

Hhollowaya also, does anyone know if memory or logging of out-of-memory are improved on the...

1984 Ford Laser•10/27/25, 11:05 PM

There's basically no reason to still be using the old KV backend

hollowaya•10/27/25, 11:11 PM

Asking because we're running into DO issues with memory exceeded (on requests that aren't using 128mb of memory), but it's not worth the migration (deleting all DO objects and remaking under a diff name) for us if the only diff is the storage change

LarryOP•10/27/25, 11:42 PM

I've never actually seen this be a problem for me (after panicking about it when I first started using DOs) but more than one instance of the same DO can share the same 128MB. Cloudflare must be smart about sensing that and not loading them that way if they are prone to use more than half the memory but do you think that might be going on?

LarryOP•10/27/25, 11:45 PM

You can sense if that is happening by storing a counter in global context. That global context is shared across instances if they are running in the same 128MB memory.

LarryOP•10/28/25, 12:44 AM

Is there a nuanced writeup comparing and contrasting extending RpcTarget vs DurableObject?

LLarry You fire and forget to offload it to another DO whose only purpose is to send fe...

LarryOP•10/28/25, 1:22 AM

As promised, we just released which offloads fetch requests to a Worker via Queues.

Introduction | Lumenize

Cost-effective external API calls from Durable Objects using Cloudflare Queues