Compute & Instances

Understanding the five plan families

CPU, RAM, disk, network profile, and price ranges across General, China Optimized, CPU Optimized, High Memory, and Storage VM families.

6 min read

Lite ships with five plan families, each tuned for a different workload shape. Every family runs on AMD silicon with DDR4 ECC REG memory and the same in-house network backbone — what differs is the CPU SKU, the storage media, the network profile, and the price-to-resource ratio.

Heads up

Plan shape is fixed at deploy time. Lite does not support resizing an existing instance — to move to a different size or family you deploy a new instance and migrate your data over. Pick the closest fit to your projected workload from the start.

At a glance

FamilyCPUStorageNetwork profileSizesMonthly price range
General (EPYC)AMD EPYC 7A23 / 7513NVMe SSDGeneraltiny → 2xlarge (6)$6 – $192
China Optimized (EPYC)AMD EPYC 7A23 / 7513NVMe SSDChina Optimizedtiny → 2xlarge (6)$6 – $192
CPU Optimized (TR)AMD ThreadRipper PRO 5975WXNVMe SSDGeneralsmall → 2xlarge (5)$18 – $288
High Memory (EPYC)AMD EPYC 7A23 / 7513NVMe SSDGeneralsmall → 3xlarge (6)$36 – $1,344
Storage VM (EPYC)AMD EPYC 7A23 / 7513HDD with NVMe cacheGeneraltiny → xlarge (5)$10 – $56
Memory is DDR4 ECC REG across every family. Disk performance follows the same fixed-policy curve regardless of family — see the Storage section.

General (EPYC)

The default starting point. Balanced CPU, RAM, NVMe disk, and bandwidth ratios that fit web servers, app backends, staging stacks, and most everyday production services. CPU runs on AMD EPYC 7A23 / 7513.

PlanvCPURAMDiskBandwidthTrafficMonthly
tiny12 GB25 GB100 Mbps2 TB$6
small24 GB40 GB200 Mbps4 TB$12
medium48 GB80 GB300 Mbps5 TB$24
large616 GB160 GB400 Mbps8 TB$48
xlarge832 GB320 GB500 Mbps16 TB$96
2xlarge1664 GB640 GB600 Mbps20 TB$192

China Optimized (EPYC)

Same EPYC hardware and per-tier price as General, but routed over the China Optimized network profile — better latency and stability for traffic terminating in mainland China. The trade-off is a smaller monthly traffic allowance per tier (e.g. 1.25 TB at "medium" vs 5 TB on General). Combine with traffic credits if your workload is bursty.

PlanvCPURAMDiskBandwidthTrafficMonthly
tiny12 GB25 GB100 Mbps500 GB$6
small24 GB40 GB200 Mbps1 TB$12
medium48 GB80 GB300 Mbps1.25 TB$24
large616 GB160 GB400 Mbps2 TB$48
xlarge832 GB320 GB500 Mbps4 TB$96
2xlarge1664 GB640 GB600 Mbps5 TB$192

CPU Optimized (Threadripper)

Backed by AMD ThreadRipper PRO 5975WX — higher per-core clocks and stronger single-thread performance than EPYC. Use this for build farms, transcoding, encoding, large analytics jobs, and anything where wall-clock CPU time is the bottleneck. Note that vCPU counts at the upper tiers (16, 32) are higher than the same-named tier in the General family for the same reason.

PlanvCPURAMDiskBandwidthTrafficMonthly
small24 GB40 GB200 Mbps4 TB$18
medium48 GB80 GB300 Mbps5 TB$36
large816 GB160 GB400 Mbps6 TB$72
xlarge1632 GB320 GB500 Mbps7 TB$144
2xlarge3264 GB640 GB600 Mbps8 TB$288

High Memory (EPYC)

Extreme RAM-to-CPU ratios for caches (Redis, Memcached), in-memory databases, large analytics datasets, and JVM workloads with big heaps. Note the deliberately small disk allocations — High Memory tiers are sized for in-memory use, not storage. The 3xlarge tier reaches 384 GB RAM on a single instance.

PlanvCPURAMDiskBandwidthTrafficMonthly
small224 GB20 GB200 Mbps5 TB$36
medium248 GB40 GB300 Mbps6 TB$72
large496 GB80 GB400 Mbps7 TB$168
xlarge8128 GB150 GB500 Mbps8 TB$336
2xlarge16256 GB300 GB600 Mbps9 TB$672
3xlarge16384 GB400 GB700 Mbps10 TB$1,344

Storage VM (EPYC + HDD)

HDD-backed bulk capacity with an NVMe cache layer in front — the cost-per-GB sweet spot for backups, archives, log retention, video/object storage, and large datasets that tolerate sequential access patterns. Pricing scales much more gently than the NVMe families: 8 TB of disk for $56/month at the xlarge tier.

PlanvCPURAMDiskBandwidthTrafficMonthly
tiny12 GB1 TB300 Mbps2 TB$10
small22 GB2 TB400 Mbps4 TB$16
medium24 GB4 TB500 Mbps6 TB$28
large24 GB6 TB600 Mbps8 TB$40
xlarge48 GB8 TB700 Mbps10 TB$56

Info

Storage VM uses HDD media — IOPS and latency are lower than NVMe families. If you need both high capacity AND high IOPS, run a NVMe-backed compute family (e.g. General or High Memory) at a larger disk size.

Decision shortcuts

  • Need a balanced default? Start with General. Almost every workload begins here.
  • CPU pegged at 80%+ all day? Move to CPU Optimized — same RAM and disk shape, faster cores.
  • Resident set growing past your RAM? Move to High Memory before you start swapping.
  • Storing TBs of cold-ish data? Storage VM saves an order of magnitude on $/GB.
  • Serving traffic primarily into mainland China? Use China Optimized regardless of compute shape.

Tip

Not sure where to start? Pick the smallest General plan, deploy your workload, and watch the live metrics for 24 hours. The dashboard makes it obvious whether you need more CPU, more RAM, or a different family entirely — and gives you a real benchmark before you commit to the production size.

Was this guide helpful?

If something is unclear or missing, our team would love to know. Open a ticket and we'll improve this page.

Contact Support