AIOps
Sustainable on-call: make workload relief change the plan
Connect on-call interruptions, specialist demand, and recovery work to a concrete capacity decision using an on-call workload worksheet.
Search articles
Search titles and article text.
How-To
Design an on-call rotation people can actually support
Define coverage, escalation, training, and recovery time before assigning the calendar. Review workload as well as shift counts.
Leadership & Culture
On-call overload: respond to the workload before it becomes normal
Use interruption patterns and responder feedback to change paging, coverage, and recovery expectations when on-call work becomes unsustainable.
AIOps
NetApp and NVIDIA: evaluate the storage path behind AI workloads
Read validated architectures as a starting point for workload, recovery, and support testing, not a promise of automatic reliability.
Leadership & Culture
Protecting time off means changing the SRE workload
Protect time away through explicit coverage, recovery policies, and realistic planning instead of relying on individual boundary-setting alone.
News
NVIDIA’s confidential inference results need a workload boundary
NVIDIA reports low confidential-computing overhead on an eight-GPU test. Separate that measured performance result from your own security and recovery claims.
News
New AI energy alliance puts workload flexibility on the SRE agenda
Google, NVIDIA and Emerald AI are backing flexible data centers. Operators will need to define which AI work can yield power, and how it recovers.
AIOps
AI in release engineering: improve evidence without weakening the gate
Use models for test suggestions and change analysis while keeping artifact identity, promotion rules, and rollback checks enforceable.
News
NVIDIA Topograph puts GPU placement evidence in focus
New Topograph guidance connects physical network topology to workload schedulers. The operational check is whether the scheduler’s view survives cluster change.
AIOps
GPU capacity depends on power and cooling
When inference slows without obvious application errors, test physical constraints alongside queueing, workload changes, and software regressions.
How-To
Linux performance tuning starts with a bottleneck
Measure a representative workload, identify the limiting resource, and test one reversible change instead of applying a universal sysctl recipe.
SRE
How blameless incident reviews make responsibility clearer
Examine what people knew and could do, then assign improvements that address the conditions behind the incident.