Linode Instances Guide

Understanding Linode Compute Instances

Linode compute instances — simply called "Linodes" — are Linux-based virtual machines that form the backbone of the Akamai Connected Cloud platform. Each Linode runs on KVM hypervisors with NVMe SSD storage, providing consistent high performance for a wide range of workloads. Unlike the complex instance type naming conventions on AWS (m5.xlarge, c6g.2xlarge) or GCP (n2-standard-4), Linode uses a straightforward naming scheme organized by plan family with clear specifications for vCPUs, RAM, storage, and monthly transfer — all at a predictable flat monthly rate.

Linode instances are available in multiple plan families, each optimized for different workload characteristics. Choosing the right plan family is the single most important decision you make when provisioning compute resources, as it determines the CPU architecture, resource allocation model, and performance profile of your workloads. This guide covers every plan family in detail, including real-world use cases, sizing strategies, and performance optimization techniques.

Whether you are running a personal blog, a high-traffic SaaS application, a machine learning training pipeline, or a real-time game server, understanding Linode's compute options helps you select the right balance of performance and cost for your specific requirements.

Plan Families Overview

Linode organizes compute instances into five plan families. Each family targets a specific class of workloads and provides a different balance of CPU, memory, storage, and network resources:

Shared CPU Plans

Shared CPU plans provide virtual CPU cores that share underlying physical CPU time with other instances on the same host. This burstable model means your Linode can temporarily use more CPU than its allocation when the host has spare capacity, but sustained high CPU usage may result in throttling. Shared plans offer the lowest price point and are ideal for workloads with variable CPU demand:

• Nanode 1 GB: 1 vCPU, 1 GB RAM, 25 GB SSD, 1 TB transfer — $5/month. Perfect for personal projects, development servers, DNS servers, and lightweight services.
• Linode 2 GB: 1 vCPU, 2 GB RAM, 50 GB SSD, 2 TB transfer — $12/month. Good for small web applications, WordPress sites, and staging environments.
• Linode 4 GB: 2 vCPU, 4 GB RAM, 80 GB SSD, 4 TB transfer — $24/month. Suitable for medium-traffic websites, small databases, and CI runners.
• Linode 8 GB and above: Scale up to 192 GB RAM with proportionally more vCPUs, storage, and transfer.

Dedicated CPU Plans

Dedicated CPU plans guarantee that your vCPUs map to dedicated physical CPU cores on the host machine. No other instance shares these cores, ensuring consistent performance regardless of what other workloads run on the same host. Dedicated plans cost more than Shared plans but eliminate the performance variability that can affect latency-sensitive applications:

• Dedicated 4 GB: 2 dedicated vCPUs, 4 GB RAM, 80 GB SSD — $36/month.
• Dedicated 8 GB: 4 dedicated vCPUs, 8 GB RAM, 160 GB SSD — $72/month.
• Dedicated 16 GB: 8 dedicated vCPUs, 16 GB RAM, 320 GB SSD — $144/month.
• Plans scale up to 64 dedicated vCPUs with 512 GB RAM for demanding workloads.

Dedicated CPU plans are the right choice for production web servers, application servers, CI/CD build agents, game servers, API backends, background job processors, and any workload where consistent response times matter. The predictable performance also makes benchmarking and capacity planning more reliable.

High Memory Plans

High Memory plans provide a higher ratio of RAM to CPU, making them ideal for memory-intensive workloads like in-memory databases, caching layers, search engines, and data analytics. These plans use dedicated CPU cores and offer significantly more RAM per dollar than standard Dedicated plans:

• High Memory 24 GB: 2 dedicated vCPUs, 24 GB RAM, 20 GB SSD — $60/month.
• High Memory 48 GB: 2 dedicated vCPUs, 48 GB RAM, 40 GB SSD — $120/month.
• High Memory 90 GB: 4 dedicated vCPUs, 90 GB RAM, 90 GB SSD — $240/month.
• Plans scale up to 300 GB RAM for the largest memory-intensive workloads.

Common use cases include Redis or Memcached caching clusters, Elasticsearch nodes, Apache Spark workers, in-memory OLAP databases, and large application caches. The relatively small SSD allocation reflects that these workloads primarily use RAM, with disk serving only for persistence and swap.

GPU Plans

GPU plans include NVIDIA RTX 6000 GPUs for machine learning, scientific computing, video transcoding, and 3D rendering workloads. Each GPU plan provides dedicated GPU cards alongside dedicated CPU cores and high memory allocations:

• GPU 1x RTX 6000: 1 GPU, 8 vCPUs, 32 GB RAM, 640 GB SSD — $1,000/month.
• GPU 2x RTX 6000: 2 GPUs, 16 vCPUs, 64 GB RAM, 1280 GB SSD — $2,000/month.
• GPU 4x RTX 6000: 4 GPUs, 24 vCPUs, 128 GB RAM, 2560 GB SSD — $4,000/month.

GPU instances come with NVIDIA drivers and CUDA toolkit pre-installed. They are available in select regions and are well-suited for model training, inference serving, video processing pipelines, computational fluid dynamics, and financial modeling. For multi-GPU training, use the larger plans or distribute across multiple GPU Linodes.

Premium Plans

Premium plans use the latest AMD EPYC processors and provide the highest single-thread performance available on Linode. They are ideal for workloads that benefit from raw CPU clock speed rather than core count, such as game servers, single-threaded applications, and latency-critical services:

• Premium plans mirror the Dedicated CPU lineup in terms of vCPU, RAM, and storage configurations but use newer-generation processors.
• Pricing carries a modest premium (typically 20-30% over standard Dedicated plans) for the improved single-core performance.
• Available in select regions that have been equipped with the latest hardware.

Creating and Managing Instances

Provisioning a Linode is straightforward through both the Cloud Manager and the CLI. Key decisions include selecting the distribution image, region, plan type, and optional features like backups, private IP, and VPC assignment:

# Create a Dedicated CPU Linode
linode-cli linodes create \
  --type g6-dedicated-4 \
  --region us-east \
  --image linode/ubuntu24.04 \
  --root_pass "SecurePass123!" \
  --label prod-app-01 \
  --tags production,app-tier \
  --private_ip true \
  --backups_enabled true \
  --booted true

# Resize a running Linode (requires reboot)
linode-cli linodes resize <linode-id> \
  --type g6-dedicated-8 \
  --allow_auto_disk_resize true

# Power operations
linode-cli linodes shutdown <linode-id>
linode-cli linodes boot <linode-id>
linode-cli linodes reboot <linode-id>

# Clone a Linode to a new instance
linode-cli linodes clone <linode-id> \
  --region us-east \
  --type g6-dedicated-4 \
  --label prod-app-02

Disk Management and Custom Images

Each Linode has a disk allocation determined by its plan. You can partition this allocation into multiple disks (e.g., a root disk and a swap disk) and manage them independently. Linode also supports custom images that you can create from existing disks or upload from external sources:

# List disks on a Linode
linode-cli linodes disks-list <linode-id>

# Create a custom image from a disk
linode-cli images create \
  --disk_id <disk-id> \
  --label "ubuntu-24-app-base" \
  --description "Ubuntu 24.04 with app dependencies pre-installed"

# Upload a custom image (requires an image file)
linode-cli images upload \
  --label "custom-os" \
  --region us-east \
  --file ./my-custom-image.img.gz

# Deploy a new Linode from a custom image
linode-cli linodes create \
  --type g6-dedicated-4 \
  --region us-east \
  --image private/<image-id> \
  --root_pass "SecurePass123!" \
  --label prod-app-03

Custom images are useful for creating golden images with pre-installed software, security configurations, and application dependencies. This reduces deployment time and ensures consistency across your fleet. Images can be deployed to any region and are stored in your account until you delete them.

Block Storage for Persistent Data

Block Storage volumes provide additional NVMe SSD storage that persists independently of your Linode instances. Volumes can be attached and detached from Linodes, making them ideal for data that must survive instance resizing, migrations, or deletion:

# Create a Block Storage volume
linode-cli volumes create \
  --label data-volume \
  --size 100 \
  --region us-east \
  --linode_id <linode-id>

# After attaching, format and mount on the Linode
mkfs.ext4 /dev/disk/by-id/scsi-0Linode_Volume_data-volume
mkdir /mnt/data
mount /dev/disk/by-id/scsi-0Linode_Volume_data-volume /mnt/data

# Add to /etc/fstab for persistent mounting
echo '/dev/disk/by-id/scsi-0Linode_Volume_data-volume /mnt/data ext4 defaults 0 2' >> /etc/fstab

Block Storage volumes range from 10 GB to 10 TB and are priced at $0.10/GB/month. They provide consistent IOPS suitable for databases, application data, and media storage. Volumes must be in the same region as the Linode they are attached to, and a volume can only be attached to one Linode at a time.

Networking Configuration

Each Linode receives a public IPv4 address, a public IPv6 address (/128 SLAAC), and an IPv6 range (/64). For private communication between Linodes in the same data center, you can enable a private IPv4 address or use VPC for more sophisticated network isolation:

# View networking configuration
linode-cli linodes ips-list <linode-id>

# Add a private IP address
linode-cli linodes ip-add <linode-id> \
  --type ipv4 \
  --public false

# Create a VPC for private networking
linode-cli vpcs create \
  --label prod-vpc \
  --region us-east \
  --subnets '[{"label": "app-subnet", "ipv4": "10.0.1.0/24"}]'

# Assign a Linode to a VPC subnet
linode-cli linodes config-update <linode-id> <config-id> \
  --interfaces '[
    {"purpose": "public"},
    {"purpose": "vpc", "subnet_id": <subnet-id>, "ipv4": {"vpc": "10.0.1.10"}}
  ]'

Performance Optimization Tips

Getting the best performance from your Linode instances requires attention to both the infrastructure choices and the software running on them:

• Choose the right plan family: Use Dedicated CPU for production workloads with consistent CPU demand. Use Shared CPU only for variable, bursty workloads. Use High Memory for caching and in-memory databases.
• Use the latest kernel: Linode lets you choose between managed kernels and distribution-provided kernels. The latest kernels include performance improvements, security fixes, and updated hardware drivers.
• Enable VPC for internal traffic: VPC networking reduces latency for internal communication compared to using public IPs and avoids counting internal traffic against your transfer allowance.
• Optimize storage I/O: Use ext4 or XFS filesystems with appropriate mount options. Consider Block Storage volumes for write-heavy workloads that benefit from dedicated I/O channels.
• Right-size your instances: Monitor CPU, memory, and disk usage with Longview and resize instances that are consistently over-provisioned or under-provisioned. Vertical scaling on Linode is straightforward but requires a brief reboot.
• Leverage transfer pooling: Network transfer is pooled across all Linodes in your account. Larger Linodes contribute more transfer to the pool, which benefits smaller instances that may occasionally spike in traffic.

Scaling Strategies

Linode supports both vertical and horizontal scaling strategies for growing applications:

• Vertical scaling: Resize your Linode to a larger plan through the Cloud Manager or CLI. Resizing requires a reboot and takes a few minutes depending on the disk size. You can resize up or down, though downsizing requires that your disk fits within the new plan's allocation.
• Horizontal scaling: Deploy multiple Linodes behind a NodeBalancer for load distribution. Use LKE (managed Kubernetes) with autoscaling node pools for automated horizontal scaling of containerized applications.
• Hybrid approach: Use Dedicated CPU Linodes for stateful services (databases, caches) that benefit from vertical scaling, and horizontally scale stateless application instances behind a NodeBalancer or Kubernetes Ingress.

For applications that need to scale beyond a single region, deploy Linodes in multiple regions and use DNS-based load balancing or Akamai's global load balancing capabilities to distribute traffic geographically.

Summary

Linode's compute offerings provide a clean, understandable set of options that cover the vast majority of cloud workloads. The transparent pricing model — flat monthly rates with hourly billing and no hidden per-request or per-GB charges — makes budgeting straightforward. By choosing the right plan family for your workload characteristics and following performance optimization best practices, you can build reliable, cost-effective infrastructure on Linode that delivers consistent results for your users.