Data Center Modernization on AWS: You Name the Exit Date, We Sign It

  • Exit date in the SOW
  • Fixed fee
  • Signed acceptance at every gate
  • Footprint decommissioned
  • AWS Premier Tier

Every data center exit starts with somebody else's calendar. A landlord's renewal notice, a vendor's end-of-support bulletin, a depreciation schedule that runs out. Mactores is the agent-native AWS modernization firm and an AWS Premier Tier Services Partner, and data center modernization is the one program we scope backwards from that date rather than forwards from a roadmap. Agents run the inventory, the mapping and the parallel-run evidence. Forward-deployed engineers own the architecture and the cutover, and they carry the date personally.

If we cause the slip, the overage cost is ours. That sentence is a clause in the statement of work, not a line in a pitch.

Get an exit date

Lease renewal on the calendar? Get an exit date.

Exit date in the SOW. Fixed fee. Signed acceptance at every gate. Footprint decommissioned.

Talk to us
AWS Premier Tier Services Partner badge
AWS Partner Tier
Premier Tier
AWS Specialization
Agentic AI
Incl. Migration & Modernization
7 Competencies
AWS Service Validations
17 Validations

60–70%

of engagement hours absorbed by agents rather than billed as analyst time

12 wks

median time to production across Mactores engagements

21

public case studies with named customers

200+

AWS-certified engineers

2008

building on AWS since — 18 years of production migrations

Mactores runs data center modernization as a dated program with one team on it from first inventory to final decommission. Nobody hands the plan to a second firm to bid. The estate is discovered once, by agents that keep the picture current instead of letting it go stale, and the same engineers who scope the exit are the ones who execute the cutover.

One team, inventory to decommission.

Discovery, target architecture, build and cutover sit with the same group. There is no re-discovery phase priced by a second vendor.

The date is a clause.

Set at signing against agreed scope. If Mactores causes the slip, Mactores carries the overage cost. Customer-caused delay converts to time and materials at published rates, visible on the invoice.

The footprint actually goes dark.

Decommission is a deliverable with a certificate attached, not a follow-on project. Contracts terminate on a schedule you approve.

Evidence, not assertions.

Every disposition carries a recorded reason. Every conversion carries a diff. Every parallel run produces output someone on your side accepted before cutover.

Top Mactores Clients

All 21 case studies →
Customers include Safaricom, Synaptics, Flipboard, Poshmark, Seagate, HP, Adani, DocuSign, KlearTrust, Tilia, Sterne Kessler and Total Expert.

Where Do Data Center Migration Programs Actually Break?

Not at the technology. At the point where a plan has to become a signature. Discovery ages out, strategy and build get bought from two different firms, and when the window finally arrives nobody has the evidence to authorize the switch. So the window moves and the old floor stays lit.

The market is tightening rather than loosening. Primary-market supply grew 33.7% year over year to a record 10,903 MW in the first half of 2026, and vacancy still fell. Under 1,500 MW remains available across all primary markets, roughly six months of demand at the current absorption rate. Renewing a footprint in that market is not a neutral act.

33.7%

Primary-market supply grew year over year to a record 10,903 MW in the first half of 2026 — and vacancy still fell.

<1,500 MW

Remains available across all primary markets, roughly six months of demand at the current absorption rate. Renewing a footprint in that market is not a neutral act.

57%

Of operators put their most recent major outage above $100,000 in the Uptime Institute 2026 outage analysis.

1 in 5

Put that outage above $1 million — the risk an unrehearsed cutover is being weighed against.

Then the program itself fails, and it fails the same three ways.
01

The inventory expires before the plan does.

A hand-built dependency map takes months. By the time it's signed off, the estate underneath it has changed, and every downstream estimate inherits the drift.

02

Two firms, two discoveries.

One writes the strategy. Another wins the build, refuses to inherit the first firm's assumptions, and re-runs the whole assessment on your money. Most programs die in that gap.

01

No one will authorize the cutover.

The date arrives. The accountable executive has status decks and no evidence the target behaves like the source. Rescheduling is the safe call, so it gets made, and the parallel run-rate continues for another quarter.

That third failure is where the money goes. In the Uptime Institute 2026 outage analysis, 57% of operators put their most recent major outage above $100,000 and one in five put it above $1 million. That is the risk an unrehearsed cutover is being weighed against.
file

Turn the plan into a signature

A date only moves the risk if somebody's balance sheet is behind it.

A scoping call produces a disposition list and a roadmap you can put in front of a CFO — not another status deck.

Book a scoping call

Why Does a Fixed Date Change the Risk on a Data Center Exit Strategy?

Because a date only means something when somebody's balance sheet is behind it. Agents carry roughly 60 to 70% of the hours a conventional proposal of equivalent scope would staff. That absorption is what lets Mactores price the work once, name a date, and put the cost of its own delay in writing.

The economics come first, because the commitment depends on them. Estate discovery, dependency and schema mapping, lineage extraction, regression generation and parallel-run comparison are the bulk of the hours in any data center exit strategy, and all of it is repeatable, checkable work. Agents do it. Forward-deployed engineers keep the judgment: which workloads retire rather than move, what the target architecture has to be, how the waves get sequenced, and when the cutover is safe to call.

60–70% Agents
FDE Judgment

Estate inventory, dependency and schema mapping, lineage extraction, regression generation, parallel-run comparison, cutover rehearsal

Disposition calls, target architecture, wave sequencing, cutover execution, alongside the phase-exit acceptance your team signs

Remove the agents and this stops being a fixed-fee business. That's the test for whether "agent-native" is a description or a marketing adjective, and it's why the phrase describes the firm rather than a service line.

What you're usually sold
What this commits to
An estimate that moves every time discovery finds something
One fee, set after assessment, written into the SOW
A target date owned by the plan rather than a person
A date in the contract, with Mactores carrying the cost of Mactores-caused delay
Steering-committee status against a milestone chart
Acceptance signed by your team at every gate before the next one opens
Strategy from one firm, build tendered separately
One team from first inventory to decommission certificate
An analyst bench billed by the hour
Roughly 60 to 70% of that layer carried by agents inside the fixed fee

The asymmetry, said once

Where the delay starts on your side, whether that is access that doesn't arrive, a decision that doesn't get made or a freeze nobody flagged, that period converts to time and materials at standard published rates. Better you read it here than discover it in a clause.

Who this isn't for

Buyers who want a large junior bench to direct. Programs where a Big 4 signature is the actual procurement requirement. Exits where the date is aspirational and everyone knows it. We'd rather say so on the first call than the third.

A date is a commitment when somebody pays for missing it. Otherwise it's a preference.

Read the commitment

The date, the fee, and who carries the cost if the date moves.

Set out in contract language you can check line by line, before any conversation about scope.

Read the commitment

What Do Data Center Modernization Services Cover on AWS?

Four workstreams and six work packages. Data platform, applications and databases, the governed AWS account structure they land in, and the decommission of what's left. Each package ends in an artefact someone on your side signs, because work you can't inspect is indistinguishable from work that didn't happen.

First, the distinction the search results won't make for you

Search this category and most of page one either defines the term or sells you a better building: power, cooling, containment, rack density, refresh cycles. That's a real discipline with real vendors, and it answers a different question. Buying a more efficient facility is not the same decision as no longer needing one. This page is about the second decision: the workloads leave, the licences stop renewing, and the floor goes back to the landlord.

Workstream
What leaves the building
Why it matters after cutover
Data platform
Warehouses, lakes, ingestion pipelines and whatever reporting has been bolted onto them
Analytics and agent workloads finally read from the system of record instead of a nightly extract of it
Applications and databases
The application tier, plus Oracle and SQL Server schemas and the stored procedures nobody has opened since the person who wrote them left
Renewal cycles end, and shipping stops being gated by a hardware order
Landing zone and governance
Account topology, network, identity, guardrails, logging, cost allocation
You get an audit position that survives questioning, and somewhere sane for everything that lands next
Exit and decommission
Whatever is left standing: residual kit, live contracts, undocumented runbooks
The recurring charges genuinely stop. This is the only workstream that reaches the P&L

Consolidating cuts the number of footprints you operate. Exiting removes them. Which one you're buying gets settled in writing at assessment, before anyone commits to a date.

The six work packages

01

Estate discovery and disposition

A live picture of every workload, dependency, data flow and licence, each carrying a 7 Rs verdict and the reasoning behind it. The crawl re-runs rather than sitting still, so the picture matches the estate on cutover day.

Artefact

A signed disposition list with a reason against every line.

02

Target architecture and landing zone

Account topology, network, identity, guardrails, logging, and the reference pattern each class of workload lands on, declared in AWS CloudFormation so the whole environment can be rebuilt or reverted as a single unit.

Artefact

A countersigned architecture and a landing zone that exists.

03

Data platform migration

Warehouses, lakes and pipelines rebuilt on AWS with lineage extracted as they move, checked by a harness comparing target output against the source under live traffic.

Artefact

Parallel-run evidence accepted domain by domain.

Data platform modernization
04

Application and database modernization

Application tier refactored, schemas converted, stored-procedure logic analyzed and rewritten, regression coverage generated continuously as the code changes.

Artefact

A regression suite your engineers own and can run with nobody from Mactores present.

Application and database modernization
05

Cutover and handover

Waves ordered by risk, the cutover rehearsed rather than merely written down, rollback criteria agreed before the window, and operational handover to whoever holds the pager on Monday.

Artefact

Customer-signed acceptance, and runbooks committed to your repository.

06

Exit and decommission

The source estate powered down, contracts and licences closed on a schedule you approve, hardware returned or disposed of under certificate.

Artefact

A decommission certificate and a closed-contract list.

What we don't do, said plainly

Anything to do with the building itself: power distribution, cooling, containment, structured cabling, physical security. Carrier contracts and circuit installation. Hardware disposal logistics past the point of certifying the decommission. End-user device refresh. Where the right answer is a specialist facilities vendor, we'll tell you which one rather than charge you while we learn their job.

How Does a Data Center Migration Plan Run, Gate by Gate?

Five phases. Each one closes at a gate your own people sign before the next opens, so the program can't drift quietly between status meetings. The fee and the date for phases two through five are committed at the exit of phase one, after discovery has seen the estate, not before.

Phase 01 Assess

Agents crawl the estate: workloads, dependencies, data flows, licences, schema complexity. Engineers turn that into a 7 Rs disposition with a reason recorded against every call, plus the target architecture and the wave sequence. This phase produces a number and a date you can put in front of a CFO, not a range.

Gate Disposition list and roadmap signed. Fee and date for phases 02–05 committed here.
Phase 02 Design

Landing zone built. Account structure, identity boundaries, guardrails, logging and residency controls go in before any workload arrives, because governance retrofitted across a live estate costs several times what it costs up front. Waves ordered by risk rather than convenience.

Gate Target architecture and SOW countersigned.
Phase 03 Build

Migration and refactoring run wave by wave against a validation harness comparing target output to source on live production behaviour. A regression surfaces the day it's introduced rather than in a test cycle three weeks later.

Gate Wave build complete, domain validation confirmed by your people rather than asserted by ours.
Phase 04 Parallel run

Target and source run together on current data for an agreed period. Rollback criteria and trigger conditions are written down before the window opens, so the call isn't made under pressure by whoever happens to be on the bridge.

Gate Parallel-run evidence accepted. Cutover window and rollback plan approved.
Phase 05 Cut over, hand over, decommission

Traffic moves. Engineers tune against real load. The source footprint is powered down, contracts terminated on your schedule, and the operational handover completes with runbooks in your repository.

Gate Customer-signed acceptance. Decommission approved and certified.

Phase 01 can be bought on its own. If the assessment concludes the exit shouldn't happen this year, that's a legitimate result and you keep the disposition list and the plan either way.

Start with the assessment

Phase 01 produces a number and a date you can put in front of a CFO, not a range.

It can be bought on its own — and if the exit shouldn't happen this year, you keep the disposition list and the plan either way.

How we work

Which AWS Migration Services Engagements Has Mactores Already Shipped?

Four, named, each with the baseline attached. Between them they cover the licence line that became the funding, a lift-and-shift that had to be rebuilt, a consolidation that closed twenty workspaces, and an audit that came back clean.

Representative engagements. Named accounts and audited figures come out under NDA.
graph

Ask for the closest reference

We will name the migration engagement nearest your estate, not the category.

Named accounts and audited figures are shared under NDA during commercial discussions.

Request a reference

How Do Data Center Consolidation Services Hold Up Under Audit?

Agent-run work is auditable when the agents produce artefacts by design rather than reports written about them afterwards. Traceability to source is a deliverable of the engagement. Internal audit will ask about this before it asks about anything else on this page.

The question that actually gets asked isn't whether the agents are careful. It's whether a human can reconstruct a decision months later. If an examiner asks why a workload was retired instead of re-hosted, the reason is in the record with a name against it.

What an examiner will ask for
Where it comes from in this engagement
Demonstrable change control over the move
Every wave carries a plan, an approval and a rollback path, each signed by a named person
Proof the new system behaves like the old one
Domain-by-domain parallel-run output, accepted by your people before the window is booked
Lineage and residency evidence
Lineage is extracted while data moves rather than reconstructed later; region and residency are guardrail-enforced in the landing zone
Separation of duties and access boundaries

Account topology, identity model and logging exist from phase 02, before a single workload arrives

Cardholder data scope (PCI-DSS)
Scope lines drawn during design. Delivered on a PCI-DSS-aligned payments platform that recorded zero audit incidents through cutover
PHI handling (HIPAA)
Designed into the target rather than remediated afterwards, on engagements that cleared internal audit first time 
Federal examination (FFIEC guidance, SEC alignment)
On regulated pricing workloads, lineage is captured at the moment of decision, with every signal, model call and outcome traced as the system runs
Independent architecture assurance
AWS Well-Architected review run against the target, with findings closed before traffic moves rather than after
Justifying a decision an agent made
Each disposition carries its reason, each conversion carries its diff, and both trace back to source

There is no separate compliance workstream to fund afterwards. The evidence set falls out of how delivery is structured, which is the only arrangement in which an agent-heavy engagement and an examiner can coexist.

  • AICPA
  • HIPAA
  • PCI DSS Compliant
  • ffiec

Audit-ready by default

Every disposition carries its reason, every conversion its diff, both traced back to source.

Bring the scope your examiner or regulator cares about and we will map it to the artefacts that produce the evidence.

Talk through your audit scope

How Does a Legacy Data Center Migration Change by Industry?

The commitment doesn't move by sector. What moves is which risk gets sequenced first, and therefore what the parallel run has to prove before anyone signs the window. A production line and a clinical system fail differently, so they leave the building in a different order.

Manufacturing

The line keeps running. SCADA and OT boundaries are mapped in phase 01, and the cutover window is fitted to the production calendar rather than the project plan. Plant operations see validation-harness evidence before they approve a window.

Manufacturing modernization

Financial Services

Evidence an examiner will accept. Lineage is produced while the data moves rather than reconstructed later, residency is settled during scoping before a single workload moves, and validation harnesses output evidence in a form a regulator can read. FFIEC and SEC alignment are design inputs here, not a review stage.

Financial services modernization

Healthcare & Life Sciences

Validation before go-live. PHI handling is drawn into the architecture and the cutover plan, using AWS HealthLake and Comprehend Medical where clinical and research workflows call for them, and research data is held to the same standard.

Healthcare modernization

Internet & Software

Waves that fit your release calendar. Migration waves are scheduled around your deploy cadence rather than against it. The strategic question underneath the exit is whether the product becomes something other systems and their agents can call, or something they route around.

Internet & Software modernization

Telco, Media, Entertainment, Gaming, and Sports (TMEGS)

Platforms with no quiet hours. Telco, media, entertainment, gaming and sport carry peak-load audiences and real-time paths. Observability is wired into the target before cutover, so peak-load behaviour is a number you can hand operations instead of a claim you make to them.

TMEGS modernization

Scoped to your sector

The commitment holds across all five. Which risk gets scoped first is yours to name.

Plant continuity, examiner evidence, PHI validation, release cadence or peak-load readiness on the line.

See all verticals

What Does a Data Center Migration Assessment Tell You About Cost?

Pricing disclaimer

Any figure, range or example in this section is directional and provided for planning context only. It is not a quote. The fixed fee for your engagement is set after assessment and confirmed in the signed statement of work.

It tells you which lines on your current cost base this program closes. Most modernization work adds a run-rate and hopes the return arrives later. A footprint exit subtracts one, which changes the funding question entirely: not where the money comes from, but which contracts end and on what date each can be served notice.

Four things move the number, and all four are settled during assessment and written into the SOW.

Cost driver
What pushes it up
Where it's fixed
Estate size and entanglement
More workloads, more integration points, more undocumented dependencies that assessment surfaces late
Locked into the SOW at the phase 01 gate, before build starts
Database estate
Larger volumes, heavier stored-procedure logic, several source engines rather than one
Locked at the same gate, priced against the disposition list
Compliance and evidence depth
Regulated data, whether PCI-DSS, HIPAA, FFIEC or SEC scope, that adds validation harnesses and sign-off steps
Scoped explicitly during assessment and priced before signature
Cutover tolerance
More dependent systems, narrower acceptable downtime, out-of-hours windows
Fixed at signing. Mactores carries the overage on delays it causes
Your own coordination
Slow approvals, delayed environment access, scope introduced mid-engagement
Converts to time and materials at standard published rates

Start with the rent, because it's the one line with a public index against it

Wholesale colocation asking rates for a 250-to-500 kW requirement reached a record $196.25 per kW per month across primary markets at the end of 2025, and rose a further 4.3% through the first half of 2026.

On that published rate, an illustrative 500 kW footprint runs roughly $1.18 million a year in rent alone, before power, before a single maintenance contract, before one hour of anyone's time.

At the increase now running through the market, renewing that same footprint unchanged costs about $50,000 more next year for exactly the capability you have today.

Substitute your own kW and your own rate. The arithmetic is shown so you can.

Where the funding is already sitting

What you're already paying for

  • Space and power for a peak that relocated to somewhere else years ago
  • A commercial database contract that renews because nobody reopens the argument
  • Extended Security Updates keeping unsupported operating systems patched
  • Refresh capital earmarked for hardware carrying half the load it was bought for
  • An integration layer whose maintenance bill exists mainly to keep the integration layer alive

What the same money buys afterwards

  • Consumption matched to real demand, with committed-use pricing once the shape settles
  • Managed engines, and one fewer renewal on the calendar
  • A supported platform, and the ESU subscription cancelled
  • Capital released for the analytics and agent work currently sitting unfunded
  • Pipelines your own team can change without raising a purchase order first

This is why the funding conversation and the engineering conversation are the same conversation. The renewal you're trying to get out of is also the date the program has to beat.

Where AWS funding fits

Mactores is MAP-eligible. The AWS Migration Acceleration Program runs across three phases (assess, mobilize, then migrate and modernize) with dedicated tracks for VMware, Microsoft, SAP and mainframe estates, and it can offset part of what assessment and migration cost.

AWS publishes its own outcome figures for migrated customers: infrastructure costs down 31% on average, infrastructure management running 62% more efficiently, and unplanned downtime cut by 69%.

Those terms are negotiated per opportunity and they move. So your AWS account manager sits in the scoping session and confirms what's live before anything enters your business case. We'll carry the paperwork. What we won't do is quote you a programme figure that expired two quarters ago.

Make the numbers yours

Put your own kW, your own rate, and the contracts you could serve notice on against the drivers above.

The assessment tells you which lines this program closes — and your AWS account manager confirms live funding in the same session.

Book a scoping call

What Terms Show Up in a Data Center Migration Services Quote?

Sixteen terms turn up in almost every proposal you'll read, and every one of them is attached to a number. The second column is what the term does to your budget, not what a dictionary says it means. Any vendor who uses one of these without volunteering the price consequence should be asked for it.

Term
Plain meaning
Budget consequence
The 7 Rs
The 7 possible verdicts on a workload: retire, retain, relocate, re-host, re-platform, refactor, repurchase
Nothing else in the estate shifts scope this much. A retired workload costs almost nothing to move; a refactored one is the most expensive line and the one that pays back longest
Re-host (lift-and-shift)
The workload arrives on AWS behaving exactly as it did before
The fastest thing to bill for and the easiest to regret, since whatever made the old bill big comes with it
Re-platform
The application stays, the engine or runtime underneath it changes
Sits between the other two on price, and is usually where a commercial licence stops renewing
Refactor
The application is restructured to work the way cloud services work
The largest build line on any proposal. Justified where the workload is the business, hard to justify anywhere else
EOSL / end of support
The day the vendor's fixes stop arriving
After it, you either buy extended coverage or you self-insure. Neither is free, and only one shows up on an invoice
ESU
Extended Security Updates, sold per year for software past its date
A subscription to stay exactly where you are, priced to get less attractive every renewal
Landing zone
The governed AWS account layout every workload lands inside
A one-off build. Skimp on it and each subsequent workload pays a remediation tax nobody quoted
Parallel run
Old system and new system processing the same live data side by side
Real money in the test phase, and the reason anyone is willing to sign the cutover. It is the first thing cut and the usual cause of a slipped date
Cutover window
The agreed hours in which traffic actually moves
Tight windows and weekend work carry a premium, and the window is set by your operations calendar rather than by the plan
RTO / RPO
Your recovery time and data-loss tolerances
Every notch tighter buys more standby capacity, and standby capacity is run-rate forever
TCO
The full cost of running something over an agreed number of years
Three to five years is the horizon where leaving a footprint wins. A twelve-month comparison almost always favours doing nothing
Egress
What a platform charges you to take your own data out
Prices the migration itself, and in a multi-cloud estate it can decide which target you pick
Colocation vs owned
Leasing space and power against holding the asset
A lease comes with a break date, and a break date is a negotiating position. An owned building brings disposal and write-down instead
AWS MGN / DMS
Application Migration Service and Database Migration Service
Tool fit decides the automated-to-manual ratio of the whole move, which is the ratio the fee is built on
AWS Control Tower
The service that governs a multi-account structure
Cheap while the accounts are empty. Expensive once there are workloads to work around
Well-Architected review
A structured assessment of the target against the AWS pillars
Remediation before cutover is a task. Remediation after cutover is a project

Which AWS Credentials Back This Data Center Modernization Work?

Everything above is Mactores' account of its own delivery. This section is not. AWS grants none of these without its own technical review, which makes them the only claims here a buyer can check independently.

Mactores is an AWS Premier Tier Services Partner, the top tier AWS awards on the services path. It also holds the AWS Agentic AI Specialization, granted specifically for agent-based delivery and a different bar from the general consulting partner status most migration firms carry.

Alongside those sit 7 AWS Consulting Competencies including Migration and Modernization, the one that governs the work described here, and 17 AWS Service Validations. The validations that carry a data center exit are the ones in daily use on these engagements: AWS Application Migration Service and Database Migration Service for the move itself, Glue and Redshift for the data platform, RDS and Aurora for the database targets, Lambda and API Gateway for the serverless layer that replaces batch, CloudFormation for the environment as a single revertible unit, and Control Tower for the account structure everything lands in. In healthcare and life sciences that list extends to HealthLake and Comprehend Medical.

aws-premier-tier-service-badge
Specialization
AWS Agentic AI Specialization
Partner tier
AWS Premier Tier Services
Governs this page
Migration and Modernization
Service validations
17 Service Validations
Team
200+ AWS-certified engineers
Building on AWS since
2008

7 Consulting Competencies

  • Migration and Modernization
  • DevOps
  • Data and Analytics
  • Machine Learning
  • AI Services
  • Healthcare
  • Manufacturing and Industrial Services

Service validations that carry a data center exit

  • Application Migration Service
  • Database Migration Service
  • Glue
  • Redshift
  • RDS
  • Aurora
  • Lambda
  • API Gateway
  • CloudFormation
  • Control Tower
  • HealthLake
  • Comprehend Medical
check

AWS credentials, verified

Every credential here is granted after an AWS technical review, so you can check it without us.

The tier, the specialization and every competency are listed on our partnership page and in the AWS Partner Solutions Finder.

See the AWS partnership

Data Center Modernization FAQs

What does a data center migration cost?
Publishing a range here would be dishonest, because every input sits inside your estate rather than ours: how many workloads, how many distinct database engines, how much of the stored-procedure logic anyone has documented, how much downtime the business will actually tolerate, how many domains need to run in parallel, and, easily forgotten, how much of the footprint can be switched off instead of moved. Assessment settles all of it. The fee and the exit date are then written into the statement of work together. Send us the footprint and the deadline; you'll hear back within one business day and have a scoped plan inside five.
What happens to our infrastructure team?
Their work changes. Whether your headcount changes is your decision and we're not going to soften that. What we will commit to is the handover: your engineers validate every domain themselves, they put their names on the gates, the runbooks end up in your repositories, and they're in the build rather than reading about it in a status report. People finish these engagements able to operate an AWS estate, which is not where most of them started.
Which workloads genuinely can't move?
A few, and assessment is where they surface, not month 7. The recurring blockers are workloads bound to specific physical hardware, licence terms that forbid cloud deployment, regulatory constraints on where data physically sits, and vendors who withdraw support the moment their software leaves your building. Each of those gets a retain or relocate verdict with the reasoning recorded, and the wave plan routes around it rather than betting somebody solves it later.
What's the rollback plan if a cutover fails?
Written and approved before the window opens, for every wave. The source system stays live and reconcilable for an agreed period after traffic moves. The rollback is rehearsed, not just documented. And the conditions that trigger it are agreed in daylight, so nobody has to invent the threshold at two in the morning with a bridge line full of people.
Do we have to freeze changes across the estate?
No. Any program asking for an estate-wide freeze is telling you its inventory has already gone stale. Freezes here are narrow: one wave, one domain, inside the window. That's only possible because the agents re-crawl the source continuously, so the mapping still describes the system on the day you move it.
Who owns the runbooks, the converted code and the IP afterwards?
You do, without qualification. Converted schemas, generated regression suites, infrastructure-as-code and operational runbooks land in your repositories under your ownership. Some of the agent tooling used during delivery is licensed rather than sold, so you are not purchasing it and nothing you receive depends on it. Everything handed over runs with nobody from Mactores in the room.
How do you price something the assessment uncovers that nobody knew about?
Surfacing those is what phase 01 is for, which is precisely why the fee and the date for the remaining phases are committed at its exit rather than at kickoff. Whatever discovery turns up before that gate gets priced into the plan you sign. Something material after it becomes a change request with its own scope, its own number and its own signature. What we won't do is absorb it quietly and hand you a slipped date instead.
What about data residency and egress cost?
Residency is enforced by guardrails in the landing zone, which makes it an architectural constraint rather than an operational promise. Egress gets sized during assessment and sits inside the migration cost, so it isn't a line that appears later. If a dataset legally cannot cross a border, that shapes the target architecture, and we would much rather learn it in week one than in the parallel run.
Is this a consolidation or a full exit?
Whichever assessment concludes and you sign for. Consolidating reduces how many footprints you operate while keeping some of them. Exiting removes them and closes the contracts behind them. Worth being exact about the difference, because only one of those two ever shows up as a reduction on your P&L.
What if we're already mid-engagement with another partner?
That happens often and it isn't an obstacle. We'll take the disposition work and architecture from wherever it stopped. We will re-run discovery, because inherited inventories are usually out of date and we're the ones putting a date in a contract. If the incumbent is delivering well and the missing piece is only the cutover, tell us and we'll scope to that shape instead.
Does AWS funding cover any of this?
Mactores is MAP-eligible. The AWS Migration Acceleration Program can offset part of the assessment and migration cost across its three phases. Terms move and are set per opportunity, so your AWS account manager joins the scoping session and confirms what's current before a single funding number goes near your business case.
faq

Ask us directly

Holding a question this page didn't answer? That is the one worth a call.

Thirty minutes with the forward-deployed engineer who would run the engagement.

Talk to us

Tell Us the Date the Lease Ends

We reply within one business day. Scoped plan inside five. A delivery lead responds, not a salesperson.
Get an exit date Email lets-talk@mactores.com
  1. 01

    Tell us the footprint and the deadline — the lease renewal, the end-of-support date, or the depreciation schedule running out. No RFP, no pre-qualification pack.

  2. 02

    You hear back within one business day, with a session with the forward-deployed engineer who would carry the exit date.

  3. 03

    You have a scoped plan inside five business days — exit date in the SOW, fixed fee — matched to your estate and compliance scope.