barklan/cto

CTO

DockerHub

cto-core | cto-porter | cto-explorer | fluentd-cto | cto-loginput

Sanity check

         │   │        │
    front│   │bot     │fluentd
         │   │        ▼
       ┌─┴───┴─┐ ┌──────────┐
┌─────►│       │ │          │
│      │porter │ │ loginput ├────┐
│ ┌────┤       │ │          │    │
│ │    └┬──┬──┬┘ └┬─────────┘   r│
│ │     │  │  │   │            ┌─┴─┐
│ │     │  └──┼───┼───────────►│pg │
│ │   rw│     │   │        rw  └─┬─┘
│ │     ▼     ▼   ▼              │
│ │  w┌───┐ ┌───────┐           r│
├─┼──►│ c │ │  mq   │            │
│ │   └───┘ └─┬───┬─┘            │
│ │     query │   │loginput      │
│ │     fanout│   │fanout        │
│ │           │   │              │
│ │ choose  ┌─┴───┴─┐ replicated │
│ │ one     │ core  │ stateful  ─┘
│ └────────►│       │ cores
│           └┬──────┘
│   callback │
└◄───────────┘

• core - meant to be replicated (one replica per node)
• loginput - can be replicated
• porter - not sure...
• mq - can be replicated through quorum queues
• pg - can be replicated cockroachdb
• c - sure, why not

cto's Issues

Badger is optimized for forward iteration

"A common configuration for Dgraph is 16 CPUs and 32 GiB of memory per machine."

https://dgraph.io/docs/deploy/production-checklist/

• [ ] Merge Log and Main storage (reserved prefix for main?)
• [ ] Switch to NVMe for ~16GB swap?