SQL Server on Linux performance

These are OS-level tuning notes for running SQL Server (Standard or Express) on Linux (VM or bare metal), including containerized deployments (Podman/Docker).

Last updated: 2025-07-02

Prerequisites

Root access on the Linux host.
tuned installed (for tuned profiles).

Tuned profile (`throughput-performance`)

Use a tuned profile designed for high-throughput server workloads:

tuned-adm profile throughput-performance

Check the active profile:

tuned-adm active
# Output: Current active profile: throughput-performance

Disable Transparent Huge Pages (THP)

SQL Server recommends disabling THP to avoid memory compaction delays that affect query performance.

Temporary (runtime only)

echo never > /sys/kernel/mm/transparent_hugepage/enabled
echo never > /sys/kernel/mm/transparent_hugepage/defrag

Persistent (systemd unit)

cat <<EOF > /etc/systemd/system/disable-thp.service
[Unit]
Description=Disable Transparent Huge Pages
After=multi-user.target

[Service]
Type=oneshot
ExecStart=/bin/sh -c "echo never > /sys/kernel/mm/transparent_hugepage/enabled && echo never > /sys/kernel/mm/transparent_hugepage/defrag"

[Install]
WantedBy=multi-user.target
EOF

systemctl daemon-reexec
systemctl enable --now disable-thp

Verify:

cat /sys/kernel/mm/transparent_hugepage/enabled
# Expected: [never]

NUMA awareness (2+ NUMA nodes)

On bare-metal or large VMs with more than one NUMA node, pin containers to reduce cross-node latency.

Check layout

lscpu | grep -i numa
numactl --hardware

Example: Pin a container to NUMA node 0

podman update --cpuset-cpus=0-5 your-sql-container
podman update --cpuset-cpus=6-7 your-api-container

I/O scheduler

Recommended scheduler for fast disks (NVMe or SSD):

cat /sys/block/sda/queue/scheduler
# Expected: [none] mq-deadline kyber bfq

none: Best for modern fast storage (default with throughput-performance)
mq-deadline: Optionally useful on SATA if contention observed

SQL Server Express limits (reminder)

When using SQL Server Express, resource limits apply automatically but you can still enforce them at the container level.

resources:
  limits:
    memory: "2Gi"
    cpu: "4"

Express Edition limits:

1.41 GB max memory

1 socket / 4 cores

10 GB per database

Checklist

Enable throughput-performance tuned profile
Disable THP persistently
Confirm I/O scheduler (none or mq-deadline)
If multi-NUMA: pin SQL/container workloads
If Express: enforce container limits when appropriate

Known risks / failure modes

Changing tuned profiles can impact other workloads on the host; validate with the full stack running.
THP settings may revert after kernel changes if not applied via systemd.
CPU pinning can make performance worse if the container threads are starved; validate with real load.

Operate & Troubleshoot Servers Services SQL Server SQL monitoring