What this work
should cost in NZ.
Almost nobody in this market publishes a rate. That makes budgeting guesswork and makes two quotes impossible to compare. These are the ranges we see pricing and reviewing specialist SQL Server work across New Zealand estates. Take them into any conversation, including one with us.
This is not a rate card. It is what a buyer should expect to pay, plus what should be inside the number before you sign anything.
Ranges are in NZ dollars and reflect what we see across NZ estates. Your quote will land where your environment puts it.
Before you approve a number, work out what the finding is worth. Licensing findings alone often recover 20 to 40% of a SQL bill.
Full assessment: expect NZ$1,000 to NZ$2,000 per instance
For that money, ask what the assessment actually runs. Expect several hundred automated checks across backups, security, patching, configuration, performance and licensing, graded, with a dollar figure and a priority against each finding. Below the range, ask what got dropped. Above it, ask what got added.
Past roughly ten instances, the per-instance number should fall
Estates get banded rates. The tooling and the reporting scale, so the marginal instance costs less to assess than the first one. If a provider quotes you the same per-instance figure for forty instances as for one, ask why.
Principal or senior DBA: expect NZ$2,000 to NZ$3,000 per day
Anything much cheaper is usually a mid-level generalist rather than a specialist. Anything much dearer should come with a named specialist and references you can call. Ask who is actually doing the work, not who is on the proposal.
The quoted price should be the price paid
Fixed-price monthly and project work is normal in this market. A meter on calls, tickets and emails is not. The only thing that should move the number is a major change of scope, agreed with you first, in writing.
The monthly spend ladder.
Managed SQL is priced per instance, per month. Each step up buys a different level of cover, not a bigger discount. Match the tier to what an outage actually costs you.
Break-fix and alignment
What this level should get you for a standard workload.
- Service desk access when something breaks
- Best-practice configuration alignment
- A workload alignment review
Nobody is watching between incidents at this price. That is the trade.
Proactive
At this level the problems should be found before they become outages.
- Everything in the tier below
- 24x7 SQL monitoring
- Automated maintenance (backups, indexes)
- Consolidated monthly reporting
- Real-time event and error alerting
If you are paying this and still finding your own problems, you are paying for the tier below.
Mission-critical
For AlwaysOn workloads where the business stops when SQL stops.
- Everything in the tier below
- HA/DR (AlwaysOn) management
- Managed failover, with a tested return to service
- An operational task bank of DBA hours
- Upgrade or migration work folded into the term
Ask for the failover test evidence. A managed failover nobody has rehearsed is a promise, not a control.
Non-production instances should be steeply discounted, at a fraction of the production rate. If a provider charges you the same for dev as for prod, push back.
What a reasonable SLA looks like.
Any provider you are considering should be able to write these three lines down without a meeting. If they cannot, that is your answer.
| Priority | What it means | First response within |
|---|---|---|
| P1 | Production down, or data at risk | Inside an hour or two |
| P2 | Degraded or at risk, still running | Same business morning |
| P3 | Advisory, planning, non-urgent | Next business day |
Read the contract for one word. These are response times, meaning how fast someone picks it up and starts work. They are not a promise to have it resolved by then. Nobody can honestly promise a resolution time for a fault they have not seen yet, and a provider who does is either guessing or padding.
Then ask who responds. A 15-minute SLA from a junior helpdesk is worth nothing if they escalate the ticket anyway. The number that matters is how long until a senior DBA has hands on the problem.
Expect these windows to apply during business hours. Round-the-clock after-hours and weekend cover is normally a separate retainer line. Check which one you are buying.
Typical project ranges.
No two environments are identical. These are the bands this work usually lands in, and what should be inside each one.
Estate discovery and roadmap
A pass over the whole data platform: what you run, where the technical debt sits, where the licensing exposure is, and a costed order of work. Ask for the roadmap to carry dollar figures, otherwise it is a wish list.
New server and Availability Group build
A new instance provisioned to Microsoft guidance, with AlwaysOn Availability Group setup and synchronisation. Ask whether a failover test and an as-built document are in the number or extra.
Version upgrade and migration
Engine upgrade end to end: compatibility assessment, testing, cutover and post-migration support. The line worth checking is the rollback plan. If there is no written rollback, the price is too low for the risk.
Something bespoke
Complex work that does not fit a band is normally priced on a senior day rate. Agree the number of days and what a day means before it starts.
What swings the price
Treat these as baselines. Four things move the final number more than anything else, and you control most of them.
- Change control. Heavy CAB oversight, narrow downtime windows and exhaustive as-built documentation can push a project up by 80%. That is a real cost of your process, not a markup.
- Light process. Nimble delivery without a documentation mountain often shaves around 20%. Say so up front and ask for it to be priced.
- Access. Waiting on credentials, VPN and third-party sign-off burns days that get billed. Have the access ready.
- Unknowns. An estate nobody has documented in five years prices higher, because the first job is finding out what is there.
Questions to ask any provider before you sign.
Five questions. Ask them of everyone on your shortlist, including us. The answers separate the quotes faster than the prices do.
Is the assessment read-only?
A read-only assessment cannot change your production instance. If the answer is anything softer than yes, ask what it writes and where.
Do you bill for calls and emails, or is it fixed-price?
A meter on communication trains you to stay quiet. Get the answer in writing before you sign.
Is the SLA a response time or a resolution time?
They are different promises. Nobody can honestly promise a resolution time for a fault they have not seen yet.
Who actually picks up a P1?
A senior DBA, or a helpdesk queue that escalates to one? Ask for the name and the after-hours path.
Do I keep the findings and scripts if we stop working together?
Your estate, your findings. If the remediation scripts leave when the provider does, you have rented the answer.
Walk away from these.
None of these are about price. They are about how you will be treated once the contract is signed.
A meter running on every email. You will stop asking, and the questions you stop asking cost the most.
A 15-minute SLA fronted by a junior helpdesk that just escalates the ticket to someone else anyway.
Pricing you can only get after three sales calls. Rates that need a relationship first are rates that vary by buyer.
A report with no dollar figures and no priority against the findings. You cannot act on a list of 400 things.
What a result should look like.
Three real engagements, kept here because they show the shape of an outcome worth paying for: a number before, a number after, and a decision someone could act on. Names are softened where a sector is narrow. The figures are exactly as they landed.
Licensed for SQL Server Enterprise, using none of its features. That is a ~$200k liability sitting on the balance sheet. Standard Edition covered the workload, and the saving was booked before the next renewal. Expect any assessment you pay for to check edition against actual feature use.
$20k to save $200k. Best buy ever.Chris, CEO
SQL deadlocks were interrupting logistics operations: orders stalling, staff re-keying, the business feeling it. Targeted work cut them by 92%. A good provider quotes the before number and the after number, not hours spent.
We can see a huge reduction in deadlocks.Deep, IT Manager
A 20TB database took up to six hours to back up against a two-hour recovery objective, so the business was quietly out of SLA and exposed on any restore. Compression brought it to 12TB and back inside the window. Any assessment worth paying for tests recovery against your stated objective.
I didn't even know this was possible.Susan, Senior DBA
What buyers ask us about pricing.
Why publish a pricing guide when providers keep rates secret?
Because making you ask is a sales tactic, not a service. If you cannot budget the work before the first call, you cannot compare two quotes either. The ranges here are what we see pricing and reviewing this work across NZ estates. Take them into any conversation, including one with us.
What should a per-instance assessment include for that money?
A read-only inspection of the instance across backups and restore testing, security, patching, configuration, performance and licensing. Ask what the assessment actually runs, and expect several hundred automated checks for this money. Ask to walk away with a graded report in plain English, a dollar figure against each finding, and a priority order. A list with no priority and no money attached is a data dump, not an assessment.
Should I pay for meetings and emails?
No. Fixed-price monthly and project work is normal in this market, and a meter on communication changes how you behave. You stop asking questions, and the questions you stop asking are the expensive ones. Agree a scope, agree a number, and agree what a change of scope looks like.
What can I get for free before spending anything?
More than most people expect. A free read-only health score shows you the shape of the risk and the overspend before you scope anything paid. Ask for the same thing from anyone you are talking to, and ask whether the findings and scripts are yours to keep.
What if the range I have been quoted sits outside these numbers?
Ask what is in the number. A quote well under the range is usually a mid-level generalist, a shorter check, or scope that gets added back later. A quote well over the range should come with a named specialist, references, and a written statement of what you get. Neither is automatically wrong. Both need explaining.
Not sure who you need?
Tell us what is going on. A senior DBA reads every message and points you at the right next step, even when that is not us. The free SQL health score is read-only, and the findings are yours to keep either way.