Any subject is accepted
The two-hour request and the two-week one follow the same path, and the difference between them shows up in the estimate. You do not have to justify size to open a ticket.
Databricks On-Demand
Resolve the data request
before it hardens into technical debt.
A broken pipeline, a dashboard nobody trusts, a wrong catalog permission, a workload cost with no owner. Send one request and get effort, timeline, and price within 24 hours. Each request is billed on its own, with no ongoing contract behind it.
Databricks On-Demand by the numbers
82h
is the global average to close a support ticket. A support contract promises response, not resolution
Geckoboard86%
of spreadsheets audited in the field contained errors. Those, running alongside your platform, are what actually carry the decision
Panko, EuSpRIG20h
per week is what controllers spend on manual data work the platform should deliver ready
Coefficient90%
success rate on small projects, against less than 10% on large ones. A focused request resolves more
CHAOS Report 2020The offer
One request at a time, on any subject of the platform.
There is no subject too small and no category out of scope, as long as it is Databricks. A pipeline that stopped, a dashboard whose number does not add up, a wrong catalog permission, a workload that got expensive, a modeling question: you describe what you need and the request enters as a numbered ticket. From there it has a written scope, an estimate, an owner and a deadline. None of this requires a support contract, a block of hours bought in advance, or a queue shared with another client.
The support plan
How Bunker delivers, and why the risk sits with Bunker.
You pay for the result validated in your own environment. The execution risk is ours. What follows is the mechanism that holds that promise up, and it is the same for every client.
Scope written before estimating
The analysis reads everything attached to the ticket, files, notes and prior exchanges, and returns the understanding in writing. If information is missing to size the work, we ask before estimating. And no estimate reaches you without passing internal approval at Bunker.
A clock that pauses when the ball is yours
The deadline runs while responsibility sits with Bunker and pauses when the ticket is waiting on you, either answering a question or signing off the proposal. The indicator measures the time that is ours.
Nothing dies in silence
A monitor flags any ticket stalled too long at any stage, and automatic reminders continue while a proposal waits for an answer. Scope, estimate, signed acceptance and audit trail stay on the ticket itself, consultable after the work is done.
Why small requests never move
Four reasons the data fix never ships
None of them is a skills problem. All of them are contracting-model problems.
The queue is shared
The request lands in a backlog serving several clients at once. Priority gets decided by negotiation, and small always loses to large.
The contract requires volume
For a two-hour fix, the vendor proposes a monthly package. You pay for capacity you do not use to get access to the one you do.
Nobody wants to sign the scope
With no clear estimate, the request sits in indefinite analysis. What was a pipeline fix grows to the size of a project, and a project needs a budget.
The knowledge left the building
Whoever built the load has left and never documented it. Touching it feels risky, so nobody does, and the manual workaround remains the process.
One request. One estimate. One owner.
Bunker
How Bunker works here
Every request stands alone. You describe what you need, get an effort, timeline, and price estimate within 24 hours, and decide whether to proceed. The estimate itself is free.
Execution has a named owner, and validation is yours, in your own environment, before any invoice.
- Member of the Databricks Consulting and SI Partner Program
- Estimate within 24 hours, no commitment
- No ongoing contract and no hour packages
- Validation in your environment before invoicing
How On-Demand works
Four phases. One request at a time. Result before payment.
The shift
What changes with one request at a time
Before
A backlog that never moves
- A two-hour fix stuck for two months
- A monthly package paid without matching usage
- Priority decided in another client queue
- A manual workaround that became the official process
- Invoicing before proof that it works
After
Requests resolved and closed
- Effort, timeline, and price within 24 hours
- Payment for the result, no idle capacity
- A named owner, no shared queue
- The fix in the platform, the manual workaround retired
- Validation in your environment before invoicing
Databricks On-Demand
Send your first request.
Describe what needs fixing or evolving in your data environment. Our team reviews the request and sends an effort estimate within 24 hours.
Every week the data request sits in the queue is a decision made on the parallel spreadsheet.
The first step is sending one request. You get the estimate within 24 hours and decide afterwards, with no upfront commitment.
See our commitment to quality: Fit to High Standards.
See also: Databricks: data and AI platform | Salesforce On-Demand | Data and AI capabilities
Frequently asked questions
Answers about Databricks on-demand
01 What is Databricks on-demand? Expand
It is one-off support on the data platform, with no ongoing contract. You describe what you need, get an effort, timeline, and price estimate within 24 hours, and execution follows acceptance. Every request is independent and billed on its own.
02 Does it work if we do not use Databricks yet? Expand
It is for teams whose environment is already running and who need a specific fix. If the platform does not exist yet, the implementation page is the right path: the foundation is designed in stages before there is anything to support.
03 Is there a minimum request size? Expand
There is no minimum. The model exists precisely for requests that do not justify a project: a stopped pipeline, a wrong permission, a metric that needs to reach the dashboard. If analysis shows the request is bigger than it looked, that appears in the estimate before you decide.