One SRE team for your cloud production.
Downtime, errors and slowness answered around the clock, and work each quarter to make them rarer.
24/7 cover and quarterly reliability work.
The flat fee covers up to 5 production services, with no licence markups or agent fees.
24/7 cover
An engineer on every alert.
- Every alert acknowledged within 15 minutes, written into your contract
- Likely cause and severity worked out within 30 minutes, also in your contract
- AI investigates every alert first, and an engineer makes the call
- Known issues fixed by runbooks you’ve approved, with every run logged
- New issues handed to your engineer with the timeline, evidence and likely cause
- A written report on the contributing factors of every incident
Reliability work
One or two goals a quarter, each with a target and a date.
- A monthly service report and a quarterly review
One identity you create from our template, read-only to start. Each runbook you approve needs one narrow permission, which you apply to that identity yourself. Revoking us is deleting it.
- AWS
IAM › Roles › tracefox › Delete- Azure
Managed Identities › tracefox › Delete- GCP
IAM & Admin › Service Accounts › tracefox › Delete
Every quarter, we work on why alerts happen.
Each quarter’s one or two goals come from these four kinds of work.
-
Reliability targets
Your product and business owners decide how reliable each service needs to be, for example 99.9% of checkouts succeeding in a month. We run the session, measure it from your users’ side and write down what you decide.
-
Alert clean-up
We page on how fast a service is using up its agreed allowance for failure, not on fixed lines like “CPU above 90%”. Quiet services get a statistical check, so a few random errors don’t wake anyone.
-
Load tests and drills
We test how production copes with heavy traffic and rehearse failures, including the runbooks themselves.
-
On‑call coaching
New issues we hand over need your engineers, so we run practice incidents with them and write runbooks and handovers that make sense at 2am.
Our method
Your business sets the reliability targets, incident reports look for contributing factors rather than someone to blame, and AI earns access in stages: it reads, then suggests, and only ever acts through runbooks you’ve approved.
From first call to steady cover.
The call and the charter cost nothing. Cover and billing start in week one, from the alerts you already have.
See how an engagement runs, step by step, with the access we hold at each one
Not included
What we don’t do.
- Fix new issues in your production by default: they go to your engineer with the evidence, unless you grant us time-limited access
- Make changes you haven’t approved
- Act as a general IT helpdesk or website support desk
Not a fit if
We’ll tell you on the call.
- Production runs on-premises. A cloud migration gets you there
- You have no alerting yet. Start with our alerts guide
- You want us to hold standing write access and fix production
- You already have three or more SREs