What SLA is and why it is in the contract
SLA is short for service level agreement. In practice it is a table of times: how many hours or business days the vendor has to give a first response to a report and how many to repair the fault. For a production plant one question matters: how long can we wait when the system stops working during a shift?
SLA is not a promise that a fault will never happen. It is a way to make the waiting time known in advance and to check it against the list of reports. That is why the InnerWeb offer describes three incident types and the availability of the server part of the instance, counted per calendar month.
Three incident types: strategic failure versus disruption
The first type is a strategic failure: the system does not perform the function on which plant operations depend. Response time is 12 hours and repair time is 8 hours from the moment service is taken up. The second type is a disruptive failure: response comes on the next business day, and the repair fits within one business day. The third type is support and planned work, meaning user questions, small configuration changes and scheduled tasks, which are completed within 7 days.
Pay attention to the words "from the moment service is taken up". Repair time counts from the moment the vendor's team starts working on the report, while response time counts from the report itself. The buyer should therefore ask which moment is recorded in the service desk system, because the report is built from it.
- type A, strategic failure: response 12 h, repair 8 h from taking up service,
- type B, disruptive failure: response on the next business day, repair within 1 business day,
- type C, support and planned work: up to 7 days.
Maximum repair time by incident type
Availability of 99.5% and how SLA is verified
Availability of the server part of the instance is 99.5% per calendar month. In a 30-day month, which has 43,200 minutes, the permitted downtime is 216 minutes, that is 3 hours and 36 minutes. The detailed method of calculation is described in a separate guide on availability.
The SLA parameters (availability, response and repair times) are set in the maintenance contract, for example availability of 99.5% per month. The basis of verification is the list of reports from the service desk system. The purchasing department should ask at the offer stage how the parameters are measured and reported, and what the parties agree if they are not met.
Service window: what it is and why updates do not go in without agreement
The service window, called the update window in the offer, is a date agreed with the client on which a new software version may be downloaded and migrated. The new version is first built and tested by the manufacturer, first on pre-production. Only then does it go to the client's instance.
For a plant running three shifts the choice of date matters in practice. The window is usually planned for the hours of lowest traffic at the gates and in permits, not on Monday morning. If the plant has peak periods, for example turnaround maintenance, the date can be moved, and the vendor does not install an update on its own initiative.
Service access outside the window: who can enter the server and when
In the on-premise model the software runs in Docker containers on the client's infrastructure as a production instance and a twin pre-production instance. The manufacturer's service access requires a personal SSH key, with no passwords, and the tunnel can be switched off by the client. Outside the update window the tunnel is disabled.
For the IT department this is the most important point in the whole SLA chapter: the vendor has no permanent entry into the plant network. Access is opened for the agreed work and closed after it is finished, and every access is assigned to a person.
What is included in the monthly fee
The monthly fee covers a dedicated instance, backups, updates and security patches, and support without billing of person-hours. The client therefore does not receive a separate invoice for each hour of service work on a report. This matters when comparing offers, because some vendors bill support by time and then the SLA becomes a paid add-on.
Video
Frequently asked questions
What is SLA in one sentence?
It is an agreed table of response and repair times for three incident types, together with server availability per month.
How is compliance with SLA verified?
On the basis of the list of reports from the service desk system, which records the times of reporting, taking up service and repair. SLA parameters such as availability of 99.5% per month and response and repair times are set in the maintenance contract.
Who sets the date of the service window?
The date is agreed with the client. The new version is first tested by the manufacturer on pre-production, and download and migration take place in the agreed window.
Can the vendor enter the server at any time?
No. Access requires a personal SSH key with no passwords, and the tunnel can be switched off by the client. Outside the update window it is disabled.
Is support billed by the hour?
No. The monthly fee includes support without billing person-hours, together with the instance, backups and security patches.
Let's talk about the terms of cooperation
We will go through the scope, SLA and purchase model with you in one meeting. Pricing is individual, and we explain the terms of the contract using examples from the plant.
Let's talk