nestor/compute · tools

What would switching providers actually take?

The bill is the easy part to compare. The hard part is what a move costs in engineering time, calendar time, and data gravity — before anything runs. Enter your setup for a transparent estimate of a do-it-yourself migration, and what the same move looks like managed.

your current setup

Workload shape
Integration depth

Default is AWS list internet egress; negotiated and PrivateLink rates differ — set yours.

estimated do-it-yourself timeline

rebuild · 3d
validate · 4d
Day 0Day 8
environment rebuild (drivers / CUDA / NCCL parity)data transferintegration reworkvalidation & re-benchmarking
38 days
calendar range (best–typical)
0.20.5 days
data transfer: 3 TB at 1.5 Gbps
13 days
environment rebuild
24 days
validation & re-benchmarking

Transfer math: 3 TB × 8,000 Gb/TB ÷ (1.5 Gbps × 86,400 s/day) = 0.2 days of pure transfer, plus overhead for retries and verification. Every input above is yours to adjust.

what a DIY migration costs — your estimates

3684 hrs
engineer-hours (your estimate)
$10,080
engineering time at $120/hr (your estimate)
$270
egress at $0.09/GB (your rate)
2
engineers pulled from product work

These are your estimates, computed from your inputs — not figures we claim about your team. The real cost is the same either way you move: engineer time diverted from your product while the migration runs.

DIY vs Nestor-managed migration

you run it

  • 2 engineers own driver/CUDA/NCCL parity, storage layout, and networking on the new provider.
  • 3 TB moved, verified, and re-permissioned by your team.
  • Re-benchmarking and burn-in on your calendar, before real work resumes.
  • Estimated 38 days and 3684 engineer-hours, by your inputs.

Nestor runs it

  • We migrate your workloads onto clusters we run for you — an extension of your SRE team, coordinating capacity across our partner network.
  • Environment parity, data movement, and cutover are our job; your engineers stay on your product instead of babysitting a migration.
  • Validation and re-benchmarking included; your team accepts the environment before billing starts.
  • For larger moves, a paid 48-hour validation run benchmarks the exact hardware before you commit anything.
  • No lock-in on the other side: root access, your own SSH keys, standard Ubuntu. If you ever leave, the environment goes with you.

The long poles are almost always environment parity and re-benchmarking, not the copy itself. That work lands on whoever owns the migration — the question is whether that is your team or ours.

take this with you

We will email you this estimate with your inputs, or copy a link to share with your team.

Want the migration scoped for real?

Timelines above are estimates from your inputs. On a scoping call we map your actual workloads, data, and cutover constraints — before any capacity is committed.

nestor/compute

Scope a deployment

Tell us what you're trying to run.

We'll follow up within one business day.

By submitting, you agree to be contacted about your request.