A 2 vCPU \/ 2 GB instance is the most common entry tier, and it is either plenty or hopeless depending on what you put on it. Below are eight workloads with the steady-state numbers I measured over a week of production traffic on the same Ubuntu 24.04 box, plus the exact commands to reproduce each measurement on your own machine.<\/p>\n\n\n\n
sudo apt -y install sysstat procps\nsudo pip3 install ps_mem --break-system-packages\nps_mem -s | head -20 # true RSS per process (shared pages counted once)\nvmstat 10 6 | awk 'NR==1 || \/^[0-9]\/'\npidstat -ru -p ALL 10 3 | egrep 'UID|nginx|php-fpm|mysqld|redis|node'<\/code><\/pre>\n\n\n\nWorkload<\/th> Steady RAM<\/th> CPU idle \/ peak<\/th> Disk<\/th> Verdict on 2 GB<\/th><\/tr><\/thead> Nginx + static site<\/td> 90 MB<\/td> 1% \/ 25%<\/td> 400 GB traffic<\/td> Easy<\/td><\/tr> WordPress + PHP-FPM + MariaDB<\/td> 620 MB<\/td> 3% \/ 85%<\/td> 2.1 GB data<\/td> Comfortable<\/td><\/tr> Node.js API + PostgreSQL<\/td> 540 MB<\/td> 5% \/ 70%<\/td> 1.4 GB data<\/td> Comfortable<\/td><\/tr> Redis as object cache<\/td> 35-210 MB<\/td> <1% \/ 30%<\/td> tiny<\/td> Easy<\/td><\/tr> Gitea + PostgreSQL<\/td> 310 MB<\/td> 2% \/ 60%<\/td> 9 GB repos<\/td> Fine<\/td><\/tr> Prometheus + Grafana<\/td> 480 MB<\/td> 4% \/ 55%<\/td> 12 GB TSDB\/90 d<\/td> Cap the retention<\/td><\/tr> MinIO object store<\/td> 260 MB<\/td> 2% \/ 40%<\/td> 70 GB bucket<\/td> Fine<\/td><\/tr> GitHub Actions runner + Docker<\/td> 700 MB<\/td> 8% \/ 100%<\/td> bursty<\/td> Tight but workable<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n1 – LEMP WordPress stack (620 MB)<\/h2>\n\n\n\n
Tuned for a 2 GB box: PHP-FPM pm = ondemand<\/code> with pm.max_children = 12<\/code>, MariaDB innodb_buffer_pool_size = 384M<\/code>, and OPcache at 128 MB. Sustained 35 requests\/s on cached pages, TTFB 180 ms, zero swap. Untuned defaults (static pool of 25 children plus a 1 GB buffer pool) OOM-killed the box within an hour of a traffic spike.<\/p>\n\n\n\n2 – Node.js API plus PostgreSQL (540 MB)<\/h2>\n\n\n\n# systemd memory ceiling so a leak cannot take the box\nMemoryMax=400M\nMemoryHigh=350M\n# postgres\nshared_buffers = 384MB\nwork_mem = 8MB\neffective_cache_size = 768MB<\/code><\/pre>\n\n\n\nWith Node pinned to 2 workers and shared_buffers=384MB<\/code>, p95 API latency stayed at 24 ms at 120 req\/s. Without the systemd ceiling, a JSON serialization leak pushed RSS to 1.6 GB in 9 hours.<\/p>\n\n\n\n3 – Redis and the 2 GB math<\/h2>\n\n\n\n
Budget roughly 100 bytes of overhead per cached key on top of the value size. One million small WordPress object-cache keys cost about 210 MB resident; a 64 MB maxmemory<\/code> with allkeys-lru<\/code> covers a typical 20k-post site. Reserve the rest for the application.<\/p>\n\n\n\n4 – Monitoring stack (480 MB, capped)<\/h2>\n\n\n\nsystemctl edit prometheus\n# [Service]\n# MemoryMax=450M\npromtool query instant http:\/\/localhost:9090 'prometheus_tsdb_head_series'\n# retention 15d -> 12 GB; 90d -> ~70 GB, too much for most 40 GB disks<\/code><\/pre>\n\n\n\n5 – A three-service Docker Compose stack (890 MB)<\/h2>\n\n\n\n
Nginx plus an application container plus MariaDB measured 890 MB resident with docker stats --no-stream<\/code>, plus about 60 MB for the daemon itself. That leaves roughly 1 GB of headroom – enough for steady traffic, but not for a container rebuild while users are active: a full --build<\/code> pinned both vCPUs at 100% for three minutes and dropped available memory to 140 MB.<\/p>\n\n\n\ndocker stats --no-stream --format \"table {{.Name}}\t{{.MemUsage}}\t{{.CPUPerc}}\"\nfree -m | head -2 # snapshot before and after compose up -d --build<\/code><\/pre>\n\n\n\nSet a mem_limit<\/code> on every service and build images in CI instead of on the production instance. That one change kept the stack out of swap during deploys on a 2 GB plan in my testing.<\/p>\n\n\n\nWhen 2 GB is not enough<\/h2>\n\n\n\n- Swap in use above 200 MB during normal traffic, not during backups.<\/li>
- PHP-FPM queue backlog (
listen queue<\/code>) non-zero in status<\/code>.<\/li>- Sustained CPU above 80% for more than 15 minutes a day.<\/li>
- MariaDB or Postgres buffer pool forced below 256 MB.<\/li>
- Two heavy workloads (for example WordPress plus a CI runner) sharing the box.<\/li><\/ul>\n\n\n\n\n
6 – Gitea plus PostgreSQL (310 MB)<\/h2>\n\n\n\n