Services
19 services entries.
-
CDN edge (hit rate + origin egress)
A CDN keeps copies of a customer's files near their users so pages load fast no matter where the visitor is. When a file is already cached nearby that's a "hit" -- served…
-
Service control plane vs. data-plane agents
A cloud service is not a billing line -- it is real software you run. It has a brain, the control plane, that decides what goes where and keeps track of everything. And it has…
-
Control-plane resilience (HA, multi-AZ)
If a service has a single brain and the server it lives on dies, the whole service loses its coordinator at once -- not just one customer, all of them. That is a single point of…
-
vCPU overcommit (the fleet dial)
You just promised more virtual CPUs than the servers physically have. That works because tenants are rarely all busy at once -- quiet boxes lend their idle cores to noisy ones.…
-
Dedicated hosts (single-tenant)
A dedicated customer wants a server all to themselves -- no other tenant's workloads on the same box. Some buy it for isolation, some for compliance, some just for peace of mind.…
-
Failover (primary -> replica promotion)
When the main database server dies, a standby copy steps in and takes over its job. There's a bumpy few minutes while the handover happens -- some requests fail -- and then…
-
Functions (per-request, per-ms billing)
With functions the customer uploads a snippet of code instead of renting a whole machine. You run it only when someone actually calls it, and you bill per request plus per…
-
GPU allocation (accelerator hosts)
GPUs are the expensive, scarce resource everyone fights over. AI labs and render studios will pay almost anything to get one -- but you only have so many cards, and a job that…
-
Load balancer (ingress + health checks)
A load balancer gives the customer one stable address -- a public IP and hostname that never changes -- and quietly spreads the incoming traffic across a pool of their backend…
-
Managed Postgres (primary + optional replica)
Plenty of customers want a fast, reliable database and absolutely none of the work of running one. With managed database you run the box -- patches, backups, the lot -- and the…
-
Managed K8s (control plane + workers)
Some customers don't want to nurse individual servers -- they want a cluster that automatically restarts their apps when a machine breaks. Managed Kubernetes gives them exactly…
-
Noisy neighbor (IOPS / CPU contention)
When several customers share one server and one of them hammers the disk, every other customer on that machine feels it. Their database queries slow down, their pages lag --…
-
Object storage (S3-style buckets)
Object storage is a place to keep files -- photos, backups, videos, app downloads -- without renting a whole server to hold them. The customer drops files into a "bucket" and…
-
Replication factor (durability tier)
Object storage keeps more than one copy of every file, spread across different machines. The replication factor is simply how many -- two copies, three copies. More copies means…
-
Scaling the control plane with the fleet
The control plane has to keep track of every server and every customer using its service. The bigger your fleet grows, the more there is to track. A tiny brain that was fine for…
-
You do not start able to sell everything. Richer services -- object storage, managed databases, CDN, Kubernetes -- begin locked. Opening one is a deliberate bet: you spend the…
-
Isolating a service with host pools
By default your services share servers -- a storage agent, a database, and a pile of customer VMs all jostle for the same CPU on the same box. That is efficient, and it is fine…
-
Uncontended compute (1:1 CPU pool)
Most tenants share CPU: you sell more virtual cores than you own and lean on the fact that they're rarely all busy at once. An uncontended customer refuses that bet. They pay for…
-
A virtual machine is a slice of one of your physical servers -- its own CPU, memory and disk -- rented to a customer as if it were a whole machine. One server runs many VMs at…