Autoscaling in 128 MiB Steps With a Spending Cap
The platform adds cloudlets (128 MiB RAM + 400 MHz CPU each) when demand rises and releases them when it drops. Set reserved cloudlets and a scaling limit. You pay for what runs, not what sits idle.
Vertical and horizontal. The platform adds CPU and RAM when load rises, adds nodes when a single container tops out.
You pay for the CPU and RAM your app consumes. Reserved resources cost less; burst resources bill only while active.
Set a reserved baseline and a scaling limit. Spending never exceeds what you allow, even during traffic spikes.
Each cloudlet is 128 MiB RAM + 400 MHz CPU. Scale in small increments instead of jumping between fixed instance sizes.
Absorb traffic spikes without over-provisioning
Set baseline reserved cloudlets, cap burst usage at the limit you choose, and let the platform add or release capacity as traffic changes.
Vertical autoscaling
Horizontal autoscaling
Scaling triggers
How billing works: RAM vs CPU
Scaling guardrails built in
-
Shared storage for stateless tiers – Externalize file uploads to shared storage so any node can serve any request. Shared storage →
-
Scale-out timing – Stateless tiers typically begin serving traffic within 30 seconds of a scale-out trigger. Stateful tiers take longer because they sync data before accepting requests.
-
Scale-in – When demand drops, extra nodes are removed so costs fall back toward your baseline.
-
Host distribution – Same-role containers run on separate physical hosts during scale-out. If one host fails, surviving nodes keep serving traffic.
-
Stop idle environments – Stop dev, staging, or batch environments between runs. Compute charges stop immediately. Only retained storage and reserved network resources continue billing.
-
Track spend in real time – The dashboard shows live resource consumption per environment. Review billing history, export CSV reports, and check cost estimates. Observability →
Set scale-out and scale-in rules in plain terms.
Autoscaling settings show CPU thresholds for adding and removing app nodes.
Monitoring starts on day one.
Alert list shows built-in load thresholds for application servers.
Start with reserved cloudlets, scale to your limit
Start a free trial with reserved cloudlets and a scaling limit already in place.
See how real workloads scale and what they cost
Use these examples as a starting point, then model exact costs in the pricing calculator.
Small API
Steady baseline, occasional spikes
One app node, 3 reserved cloudlets (384 MiB), scaling limit of 16. Handles steady REST traffic and absorbs occasional spikes without a second node.
App Hosting overviewWordPress
Single-node or clustered
Single-node WordPress: 3 reserved cloudlets, 16 max. Clustered WordPress: 2 app nodes (2 reserved / 8 max each), 2 DB nodes, and a load balancer. Horizontal scaling starts when both app nodes top out.
WordPress hostingJava cluster
Tomcat / GlassFish with load balancer
Tomcat or GlassFish behind a load balancer. The Tomcat preset runs 2 app nodes with 0 reserved and 6 max cloudlets each, so idle periods bill only for disk and IP fees.
Java hostingNode.js
Per-service scaling limits
Run services in separate environments with independent limits. Give the payment API a higher scaling limit; keep the admin dashboard low. Each environment is billed independently.
Node.js hostingDatabase HA
Replication presets compared
MySQL primary-secondary: 2 nodes, 4 reserved / 16 max each. Galera multi-primary: 3 nodes, 1 reserved / 16 max each. PostgreSQL: 2 nodes, 0 reserved / 32 max. With zero reserved cloudlets, idle periods cost only disk and IP fees.
Database clustersKubernetes
Worker-node autoscaling
Worker nodes start at 5 reserved / 32 max cloudlets. The control plane runs on platform-managed resources. Add workers horizontally as pods scale out.
Kubernetes hostingPrefer fixed instances? See PaaS vs VPS. Want usage history and cost alerts? See Observability.
Common Questions
How does pricing work?
App Hosting is billed hourly from a prepaid balance. You set your reserved baseline and burst limit, and autoscaling stays within those limits.
How is this different from VPS hosting?
App Hosting gives you prebuilt stack templates and scaling controls, so you do not have to assemble and operate each layer yourself. If you need full OS control, Cloud VPS is the better fit.
What is a cloudlet?
A cloudlet is one unit of compute: 128 MiB RAM + 400 MHz CPU. You set baseline and burst scaling in cloudlets.
Can I scale without downtime?
Scaling adds capacity within the limits you set. Whether that happens without downtime depends on your architecture and application. Start with a calculator preset, or chat with an engineer about the right setup.
Do you support HA databases?
Database Clusters covers MySQL, MariaDB, and PostgreSQL with HA options.
Is there Kubernetes support?
Kubernetes Hosting is available, with architecture presets to get started.
Can I run production on this?
Many customers run production on App Hosting. Use the trial for hands-on evaluation, and talk to an engineer to design your production setup.
Can you help me migrate?
We can map a target setup and migration plan. Portability & Migration has the details.
Is this a managed service?
Togglebox manages the platform and infrastructure, including the runtime, networking, and base OS layers. Your team manages the application code, dependencies, and runtime upgrades.
Can I upgrade or downgrade runtime versions?
You choose when to switch runtime versions (PHP, Node.js, Java, Python, Ruby, Go, .NET). Test in staging, then promote to production or roll back in a few clicks.
What happens when traffic spikes past my baseline?
The platform adds cloudlets up to the maximum you set. In most cases, scaling happens within seconds. When traffic drops, resources scale back down and billing follows.
What happens when a node goes down?
In a clustered setup, the load balancer routes traffic to the remaining healthy app nodes. For databases, replication modes such as primary/replica and Galera help keep data available during a node failure.
What happens if I set limits too low?
Your app uses all available cloudlets up to your maximum, then stops scaling. It doesn't crash. You get notified and can raise the limit from the dashboard.
What happens to charges when I stop an environment?
RAM, CPU, and traffic charges stop immediately. Charges for retained storage and reserved resources, such as public IPs and SSL certificates, continue. That lets you stop idle dev or staging environments while keeping data and network resources in place.
What deployment methods are supported?
Deploy with application archives (WAR, JAR, ZIP, or EAR), Git or SVN repositories, URL-based archive pulls, or build-node workflows. You can also add pre- and post-deploy hooks for smoke tests, migrations, or cache warming.
Can multiple team members share access to environments?
Invite collaborators with role-based permissions. All usage bills to the primary account.
Do you support two-factor authentication?
Protect platform access with 2FA using time-based codes from an authenticator app. Recovery codes provide backup access.
Can I clone an environment for testing or staging?
Clone full environments in minutes for testing, staging, release rehearsal, or A/B scenarios. Cloned environments include the full setup and configuration.
Is Memcached available for caching?
Add Memcached for distributed in-memory caching to reduce database load and improve response times. It also supports session storage for PHP and Java apps in clustered setups to maintain session continuity during failover.
Can I automate deployments with pre- and post-deploy hooks?
Run pre- and post-deploy hooks for tasks like smoke tests, database migrations, and cache warmups. Automation & CI/CD has more examples.
No matching questions found.
Start your 14-day free trial
Deploy in minutes with managed autoscaling and clustering built in.
No credit card required.