Social Wire
Tweeted
Aug 26, 2025
VEXXHOST, Inc.
@vexxhost If provisioning a GPU node takes longer than a training epoch, your cloud’s the bottleneck. A reproducible stack: - @OpenStack+K8s - GPU passthrough/MIG - DCGM+Prometheus - Ceph for datasets & checkpoints - OVN/DPDK for low-latency And no lock-in.vexxhost.com/blog/open-sour…3z
VEXXHOST, Inc.
@vexxhost If provisioning a GPU node takes longer than a training epoch, your cloud’s the bottleneck. A reproducible stack: - @OpenStack+K8s - GPU passthrough/MIG - DCGM+Prometheus - Ceph for datasets & checkpoints - OVN/DPDK for low-latency And no lock-in.vexxhost.com/blog/open-sour…3z
Tweeted
Aug 26, 2025
VEXXHOST, Inc.
@vexxhost Prod in minutes and provable controls: Keystone/Keycloak RBAC, Neutron isolation, IaC+GitOps, Barbican secrets, CI/CD ephemeral envs, Octavia blue-green, Prometheus gates, Cinder/Velero rollbacks. Here's how to wire secure DevOps on @OpenStack (via Atmosphere):
VEXXHOST, Inc.
@vexxhost Prod in minutes and provable controls: Keystone/Keycloak RBAC, Neutron isolation, IaC+GitOps, Barbican secrets, CI/CD ephemeral envs, Octavia blue-green, Prometheus gates, Cinder/Velero rollbacks. Here's how to wire secure DevOps on @OpenStack (via Atmosphere):
Tweeted
Aug 25, 2025
VEXXHOST, Inc.
@vexxhost Local NVMe is fast; it’s also a single-host blast radius. 🫤 The blog shows you how to scale local storage on OpenStavexxhost.com/blog/how-atmos…NxJiZ
VEXXHOST, Inc.
@vexxhost Local NVMe is fast; it’s also a single-host blast radius. 🫤 The blog shows you how to scale local storage on OpenStavexxhost.com/blog/how-atmos…NxJiZ
Tweeted
Aug 22, 2025
VEXXHOST, Inc.
@vexxhost #Kubernetes, simplified. #Atmosphere + #Magnum Cluster API = automated clusters, seamless scaling, and resilient lifecycle management on @OpenStack.
VEXXHOST, Inc.
@vexxhost #Kubernetes, simplified. #Atmosphere + #Magnum Cluster API = automated clusters, seamless scaling, and resilient lifecycle management on @OpenStack.
Tweeted
Aug 21, 2025
VEXXHOST, Inc.
@vexxhost Hadoop on VMs, Spark/TF on K8s, GPUs for training, streams to ingest. Atmosphere aligns VMs + @kubernetesiovexxhost.com/blog/how-atmos…AYN
VEXXHOST, Inc.
@vexxhost Hadoop on VMs, Spark/TF on K8s, GPUs for training, streams to ingest. Atmosphere aligns VMs + @kubernetesiovexxhost.com/blog/how-atmos…AYN
Tweeted
Aug 20, 2025
VEXXHOST, Inc.
@vexxhost Scaling SaaS hits limits at tenancy boundaries, network edges, and cluster lifecycle (and not at CPU graphs). We share failure modes + reference patterns on @OpenStack - isolation that sticks, versioned clusters, signals-driven scaling, sane storage. Read:
VEXXHOST, Inc.
@vexxhost Scaling SaaS hits limits at tenancy boundaries, network edges, and cluster lifecycle (and not at CPU graphs). We share failure modes + reference patterns on @OpenStack - isolation that sticks, versioned clusters, signals-driven scaling, sane storage. Read:
Tweeted
Aug 19, 2025
VEXXHOST, Inc.
@vexxhost A sane mix: baseline with declarative, cross-app glue via iPaaS, business workflows in low-code, bespoke gaps in scripts - behind one IAM, one secrets store, one audit trail. What’s your split?
VEXXHOST, Inc.
@vexxhost A sane mix: baseline with declarative, cross-app glue via iPaaS, business workflows in low-code, bespoke gaps in scripts - behind one IAM, one secrets store, one audit trail. What’s your split?
Tweeted
Aug 19, 2025
VEXXHOST, Inc.
@vexxhost 4. Declarative: desired state for infra/config (Terraform, Ansible modules). Plan/apply, drift detection, inventories, and rollback artifacts make fleet changes repeatable.
VEXXHOST, Inc.
@vexxhost 4. Declarative: desired state for infra/config (Terraform, Ansible modules). Plan/apply, drift detection, inventories, and rollback artifacts make fleet changes repeatable.
Tweeted
Aug 19, 2025
VEXXHOST, Inc.
@vexxhost 3. iPaaS: cross-system event bus with managed connectors. Check delivery semantics (at-least-once), DLQs/replay, quotas, and data residency. Offload heavy compute to workers.
VEXXHOST, Inc.
@vexxhost 3. iPaaS: cross-system event bus with managed connectors. Check delivery semantics (at-least-once), DLQs/replay, quotas, and data residency. Offload heavy compute to workers.
Tweeted
Aug 19, 2025
VEXXHOST, Inc.
@vexxhost 2. Scripting: perfect at the edges (adapters, transforms). Keep it idempotent, packaged, logged.
VEXXHOST, Inc.
@vexxhost 2. Scripting: perfect at the edges (adapters, transforms). Keep it idempotent, packaged, logged.