Grafana
Each of these is a real decision Grafana is working through, written so you can make the call yourself — incomplete information, someone who pushes back, no answer key.
Grafana Labs provides open-source and cloud-based observability software for collecting, querying, visualizing, and alerting on metrics, logs, traces, and profiles. It serves developers, DevOps, SRE, and platform teams running cloud-native or complex distributed systems, through open-source, managed SaaS, and enterprise self-managed options. Grafana Cloud lets customers use open source observability software including Prometheus, Mimir, Loki, Tempo, and Pyroscope without installing, maintaining and scaling their own observability stack. The telemetry databases are Mimir for metrics, Loki for logs, Tempo for traces, and Pyroscope for profiles. They are offered as a hosted service in Grafana Cloud and as on-premise solutions with Grafana Enterprise Metrics, Grafana Enterprise Logs, and Grafana Enterprise Traces. They are multi-tenant distributed systems implemented in Go and running on Kubernetes across all major Cloud service providers (GCP, Azure, AWS). Grafana Enterprise is a composable observability platform focused on large-scale operators with security and regulatory requirements who need to self-operate their observability solutions.
Your first one is free. No card.
25 live right now.
The rest
Route SLO-burn alerts with evidence before bridge calls stall
Tenant SLO-burn alerts must reach a reviewed severity decision in under 10 minutes for 80% of the pilot. You must weigh faster routing against evidence quality and the engineer’s authority to decide.
Trace alert failures while engineers still verify every cause
Alert-evaluation investigations must produce engineer-confirmed causes within 20 minutes in 70% of pilot cases. You have to weigh faster evidence assembly against the review needed before a cause guides a fix.
Draft qualification fields from calls while sellers approve edits
Growth sellers are revisiting call records because key qualification details are thin or unsupported in Salesforce. You must weigh faster usable next steps against preserving seller review and source evidence.
Settle who owns a failed run before the pod disappears
Four customer workshops surfaced the same gap: failed Kubernetes jobs emit metrics but lack a consistent owner and investigation view. You must weigh a simple deployment kit against the context SREs need after short-lived workloads disappear.
Prove critical metrics before the one-way cutover
Nineteen of 240 critical metric series are already absent, and another 27 are only machine-flagged. You must weigh the approved collection boundary against the evidence needed to trust incident response after cutover.
Route pages to owners while the label freeze holds
Twenty-seven percent of the customer's alerts now land in a shared queue, stretching acknowledgement to 18 minutes. You must weigh a reversible routing pattern you can roll back against the durability of fixing ownership at the workload source.
Test a one-market offer for 20 teams while Finance caps exposure
High-volume teams are comparing managed Mimir, Loki, and Tempo with operating those components themselves. You must weigh a compelling entry offer against commercial boundaries that can support a wider launch.
Settle a forecast for regulated accounts while evidence conflicts
Automated analysis projects a 22% contraction, but direct records confirm it for only 58% of affected accounts. You must weigh capacity protection against a forecast assumption that will be expensive to reverse.
Fix the cohort view so flagged accounts match criteria, not just 63%
Expansion conversion fell from 18% to 9% despite 31% segment consumption growth. You must weigh a cleaner cohort signal against continuity in the report leaders already use.
Expose token-heavy paths without inflating telemetry costs
SREs can see rising token spend but still rebuild request evidence outside their telemetry workflow. You must weigh investigation speed against tenant-safe, affordable data retention.
Choose a session-data path while replay evidence disappears
Some session replays omit the events SREs need during active investigations. You must weigh immediate investigation access against a storage decision that cannot be cheaply undone.
Stop doubled logs from new clusters instead of patching searches in Loki
Eighteen percent of newly enrolled clusters are sending duplicate container logs into Loki, and your fix decides whether those users can trust their incident searches. You must protect existing collector deployments while restoring logs that SREs can rely on.
Win evaluations with SRE leaders across clouds before bundled tools land
Cloud-provider sellers are introducing bundled observability tools before multi-cloud SRE teams reach an evaluation of your product. You must weigh partner reach against a motion that still gives each field team a reason to engage.
Hold the expansion while self-hosting enters the budget review
A $1.5 million expansion is stalled because finance rejects uncertain AI-feature charges while SREs need broader incident coverage. You must weigh a defensible spend boundary against the access the reliability organization needs.
Close the trial before procurement questions the managed-service value
The proof of concept ends in 14 days while procurement still cannot connect its observed incident gains to the quoted spend. You must weigh a faster close against a value case the SRE lead and economic buyer will both defend.
Move teams from setup into documented investigations within 90 days
Customers enable Pyroscope on a few services, then stop before it becomes an investigation habit. You need to weigh a focused customer programme against a broader learning design that serves more future accounts.
Commit the renewal path before self-operation gets funded
Loki volume is up 46% quarter over quarter, and projections place spend above the approved ceiling. You need to weigh a durable managed adoption case against a self-operated path that gives the customer more direct control.
Fix the stale state of alert links while releases keep shipping
Only 42% of alert-linked sessions reach a useful dashboard state, down from 79%. You need to weigh a fast configuration change against an on-call flow that stays clear across future releases.
Choose a path for each workload so SREs stop abandoning setup
Users abandon setup after comparing SDK, Beyla, and Collector paths with no trusted recommendation. You must weigh broad workload coverage against a first version that users and coding agents can verify.
Decide on an attribution gate without blocking healthy ingestion
Twenty-nine percent of affected Kubernetes workloads have conflicting service attribution. You must weigh immediate telemetry continuity against a cutover that makes downstream diagnosis trustworthy.
Fix Java onboarding so first-attempt telemetry returns to 78% from 61%
Java SDK completion fell from 78% to 61% after the onboarding sequence changed. You must weigh a faster default path against the flexibility advanced users need.
Give tenants a way into telemetry when identity providers fail
Identity disruption has delayed access to incident telemetry for high-SLA customers. You must weigh recovery speed against tenant isolation and least-privilege controls.
Stop a suspected leak in Tempo without losing incident queries
Two sampled queries suggest trace attributes may have crossed tenant boundaries. You must weigh broad containment against the loss of trace visibility during active customer incidents.
Trim Mimir burn during the daily reporting window
Eight reporting windows have pushed one tenant to 38% of its monthly error budget. You must weigh a quick reliability gain against alert noise and safe production change.