Container Compute
Send an image. Get a URL back.
Container Compute runs your app from a container (in Thai) image on servers in Bangkok 1 and gives it an HTTPS URL. Instances are added as requests arrive, and the service scales down to zero when none do. You pay for the virtual CPU (vCPU), memory and requests you use.
- Compute cost at zero instances
- ฿0
- Free vCPU-seconds per month, per container
- 180,000
- Supported images
- linux/amd64
- URL on *.run.xver.cloud
- HTTPS
Four steps from image to URL
- Your image
- Build
- Deploy
- HTTPS URL
-
Send an image or connect GitHub
Use an image from Docker Hub, GHCR or your Artifact Registry, or connect a repo through the XVer GitHub App and let it build the image.
-
Set vCPU, memory and concurrency
Choose vCPU and memory per instance, how many requests one instance handles at once, and the time limit per request.
-
Pass the health check, go live
The new version must pass a health check before it receives any requests.
-
Get an HTTPS URL
The service gets an address on *.run.xver.cloud with HTTPS.
Autoscaling and scale to zero
When requests exceed what one instance takes at once, another instance starts. When requests drop off, instances stop until none are left.
-
Scale to zero
While no requests arrive, no instance runs, so there is no vCPU or memory charge.
-
Minimum instances and cold starts
After scaling to zero, the first request waits while an instance starts. This is a cold start. To always answer right away, set a minimum of 1 instance or more. That instance is billed for vCPU and memory for as long as it runs.
-
Request-based or instance-based
Request-based adds and removes instances with the request count and can go down to zero. Instance-based keeps instances running and bills for the time they run. Both use the same rates.
Example configuration
Instances
Instances over one day for an online shop. No one visits at night, so it scales down to zero.
- Running instances (minimum 0)
- With a minimum of 1 instance, it never drops below this line
Workloads that suit Container Compute
-
LINE or Discord bot webhooks
- Problem
- The bot only gets messages now and then, yet it needs a server waiting 24 hours a day.
- What you get
- Paste the service’s HTTPS URL into the webhook settings. While no messages arrive, the service sits at zero instances.
-
Backend API for an online shop
- Problem
- Sales bring a rush of visitors, but nights are nearly empty. Renting a big server for the peak means overpaying most of the month.
- What you get
- Instances are added as visitors arrive and removed when it goes quiet. You pay only for the instances running at each moment.
-
Trying new versions from GitHub
- Problem
- You want your team or a client to try a new version before it goes live, without leaving a test server running.
- What you get
- Create a separate container for the test version from the same repo. It gets its own URL and scales to zero when no one is looking.
-
Background and scheduled jobs
- Problem
- Some work runs on a schedule, like a nightly sales summary, or pulls tasks from a queue, but keeping a server on all day is not worth it.
- What you get
- Let an outside scheduler such as GitHub Actions call the job URL on a schedule, and pay only while the job runs. Work that watches a queue all the time uses instance-based mode.
Container Compute pricing
Usage is metered every second, and every hour the charge is deducted from your prepaid credit in Thai baht. The free tier resets on the 1st of each month.
Rates
- vCPU
- ฿0.0003 /vCPU-second
- Memory
- ฿0.00005 /GiB-second
- HTTP requests
- ฿20 /1M requests
Free tier per container, per month
- 180,000 vCPU-seconds
- 360,000 GiB-seconds
- 1,000,000 requests
| Item and how it adds up | Cost |
|---|---|
| vCPU 270,000 − 180,000 free = 90,000 vCPU-seconds × ฿0.0003 | ฿27.00 |
| Memory 540,000 − 360,000 free = 180,000 GiB-seconds × ฿0.00005 | ฿9.00 |
| HTTP requests 500,000 requests, within the 1,000,000 free | ฿0.00 |
| Total per month | ฿36.00 |
At this size, about 200 hours a month fit entirely in the free tier. This example leaves out Block Storage, which is billed separately.
Container Compute or Cloud Compute
| Container Compute | Cloud Compute | |
|---|---|---|
| Best for | Apps used in bursts, or new apps where you do not know the traffic yet | Work that runs steadily 24 hours a day |
| Billing | The vCPU, memory and requests you use, metered per second | Hourly per plan, never more than the monthly price |
| When no one uses it | Scales to zero, no compute charge | Billed from creation until you delete the server, even while it is shut down. Delete it to stop charges |
| When traffic grows | Instances are added automatically | Size set by the plan. A bigger plan means a resize, and the server restarts |
| What you look after | Just your image or app code | The whole server, from the operating system up |
| Example: on 24/7 at 1 vCPU · 1 GiB | About ฿847.80/mo (excluding requests) | ฿180/mo (plan s-1vcpu-1gb) |
Services to use with Container Compute
-
Block Storage
Mount a volume into a container at the path you choose. It can grow while your app runs, but it cannot shrink.
A container with a volume runs as a single instance.
฿4 /GB-month
-
Artifact Registry
A private store for container images inside your project. Push an image and pick it in Container Compute right away.
The first 0.5 GB are free. Storage above that is metered hourly, per project.
฿1.50 /GB-hour
Questions about Container Compute
Do I need a Dockerfile?
No. If you already have an image, use it as is. If your code is on GitHub without a Dockerfile, we build an image for you with buildpacks for Node, Python, Go and JVM.
You can also write a Dockerfile straight into the settings page.
Can I build images on an Apple Silicon Mac?
Yes. Container Compute runs images built for linux/amd64, so build with docker build --platform linux/amd64.
Is data inside a container lost on redeploy?
Files your app writes inside the container are not permanent. To keep data, mount a Block Storage volume. Data on the volume stays through redeploys, restarts and image changes.
Start with Container Compute
Sign in with GitHub, then top up credit in Thai baht with PromptPay, TrueMoney Wallet, Visa · Mastercard · JCB cards, or Internet Banking. Support answers 24/7 in Thai and English.