Intelligent time series database migrations

Unblock your database migration before kickoff.

Tamae maps what your estate contains against what the software needs. You share a machine-readable version at kickoff, so all parties work from your estate instead of asking you to describe it.

The problem

Scrambling to work out what the software needs from your estate stalls the implementation.

What the software needs isn’t written down in one place. It gets assembled by hand, mid-project, by people who are already busy — and what was missing only shows up once the work has started.

30% longer than planned, on average
65% exceed the budget they set
<10% finish on the original schedule

Horváth (2025), 200 enterprise platform transformations — respondents named underestimated data migration among the main causes.

There’s no single artefact that explains your setup

The picture of your estate is spread across systems and half of it sits in people’s heads. Whatever was written down describes how things were, not how they are.

You don’t know exactly what the software needs

So what you send is incomplete in ways neither side can see. Both parties work from it anyway, until something doesn’t line up.

It surfaces after kickoff

The same missing detail costs days before kickoff and months after. Found late enough, it blocks the project outright.

Both sides carry it

You can’t commit to a date you can’t scope. The vendor can’t staff or sequence delivery, and can’t get you live and seeing value.

How Tamae works

From an estate nobody can share to a map your vendor can work from.

Connect your source systems. Tamae builds the picture, keeps it current as the estate moves, and hands all teams something machine-readable at kickoff.

01

Connect your source systems

Point Tamae at the estate — schemas, on-disk layout, volumes, query history. Read-only, and it stays in your environment.

Source estate
Schemas On-disk layout Query history
Target platform
Type system Partitioning rules Capacity limits
Read-only · 412 tables profiled, nothing leaves your environment
02

Tamae builds the picture

Every table and column mapped against what the target platform actually needs from it. Your team reviews before anything is confirmed.

Proposed for the map3 sources read
schema-catalog412 tablespartition-layout96 partitionsquery-log-sample240 queries
Your team confirms before anything enters the map
03

A map that changes when the estate does

The systems keep moving while the project runs. The map moves with them, so scoping done in month one is still true in month four.

Live mapupdated 12m agoupdated just now
mappedtrade.exch_timedesignated timestamp, ns precision
mappedflaggedorder.venue_idstable id, unique per venueno issues foundregenerated per session — 18,400 ids
3 sources in the mapquery-log-sample just added
04

Exceptions surfaced and routed

What has no equivalent in the target, what replays, what needs a decision — each with an owner against it, before it becomes a blocker.

AGENTlast run · 14 min ago
  • Reconciled 3,908 columns against the target type system
  • Flagged 11 with no equivalent in the target
  • Routed each one to the team that owns it
11 exceptions routed since the last check-in
05

Share it at kickoff

All teams get a machine-readable version to work from directly, instead of interpreting a document or asking you to describe your setup. Export it, or share a password-gated link.

What’s still open before we can commit to a date?
Eleven columns. quote.cond_code has no equivalent type in the target — three variants across 41M rows, market data team. order.venue_id regenerates per session; the platform vendor has a fix drafted.
Live link · password-gated Machine-readable export
What you get

What a discovery phase doesn’t leave behind.

Built before the contract

The mapping happens during the sales cycle, so you reach kickoff ready instead of finding out at kickoff.

Machine-readable, not a document

All teams work from the map directly. Nobody re-keys it, nobody interprets it, nothing is lost in the reading.

Stays in your environment

Tamae reads the estate from your side. What crosses to the vendor is the mapping, and you decide when it does.

We do the mapping, not the build

The loader, the platform tuning, the cutover and the sign-off stay with your team and the engineers.

Current, not a snapshot

The map tracks the estate as it changes, so scoping doesn’t go stale halfway through the project.

It outlasts the programme

The next migration starts from a current picture of the estate, across platforms and across projects, instead of building one again from scratch.

For database vendors and delivery partners

Deals stall on client readiness. You’re not staffed to fix that.

Unblocks the deal, not just the delivery

The client arrives at kickoff with a mapped estate, so scoping stops being the thing holding up the contract.

No services function to build

The repeatable mapping layer, done once — instead of your implementation team rebuilding it differently on every engagement.

Faster to live

Clients who reach production quickly expand. The ones stuck in implementation churn.

Before kickoff

What exactly does your new platform need from your estate?