View server resource usage from inside the DB (CPU/IO)

Oracle

SELECT stat_name, value FROM v$osstat
WHERE stat_name IN ('NUM_CPUS','BUSY_TIME','IDLE_TIME','IOWAIT_TIME','PHYSICAL_MEMORY_BYTES');
-- Real-time metrics: SELECT metric_name, value FROM v$sysmetric WHERE group_id = 2;

v$osstat gives OS CPU/memory counters; v$sysmetric gives real-time DB metrics (CPU, IOPS) for the latest interval.

PostgreSQL

SELECT backend_type, object, context, reads, writes, hits, evictions
FROM pg_stat_io
ORDER BY reads DESC;
-- Note: CPU usage is not visible from inside the DB; an OS tool (top, sar) is required.

pg_stat_io gives read/write/cache hit counters; PostgreSQL does not expose CPU usage, so operating system tools are required.

MySQL

SELECT * FROM sys.host_summary;
-- For InnoDB I/O and thread details:
SHOW ENGINE INNODB STATUS\G

The sys schema summary views (host_summary) give an IO/latency breakdown; SHOW ENGINE INNODB STATUS shows buffer pool and IO activity.

SQL Server

SELECT object_name, counter_name, cntr_value
FROM sys.dm_os_performance_counters
WHERE counter_name LIKE '%CPU%' OR counter_name LIKE '%Page life expectancy%';
-- CPU history: sys.dm_os_ring_buffers (RING_BUFFER_SCHEDULER_MONITOR)

sys.dm_os_performance_counters gives DB performance counters; dm_os_ring_buffers gives the SQL/other process CPU ratio for the last ~256 minutes.

MongoDB

db.serverStatus().metrics
// Relevant fields: .opcounters, .network, .wiredTiger.cache, .mem, .extra_info.page_faults

serverStatus() gives memory, network, opcounters, and WiredTiger cache metrics; for detailed OS CPU/IO use mongostat or operating system tools.

Redis

INFO cpu
INFO stats
# used_cpu_sys, used_cpu_user, instantaneous_ops_per_sec, total_net_input_bytes

INFO cpu reports the Redis process's CPU consumption, and INFO stats reports per-second operations and network traffic.

Elasticsearch

GET _nodes/stats/os,process
// Look at the os.cpu.percent, os.mem, process.cpu.percent, fs (disk) fields

_nodes/stats/os,process gives CPU percentage, memory, and process metrics per node; use the fs section for disk.