What Is Await In SAR Output?

Q: How can I use iostat to dig deeper?

Run iostat -x 1 3 to pull extended disk stats, including await, %util, and avgqu-sz. This gives you a clearer picture of whether the slowdown is disk-bound or queue-driven. Picture a %util near 100% alongside a sky-high await —that screams a saturated disk.

Q: Where do I look for disk errors?

Check dmesg | grep -i error or /var/log/messages for disk-related errors. Failing disks often push await through the roof. If you spot errors, swap the drive immediately—hardware issues won’t fix themselves.

Q: Can a single process cause high await?

Yes—use sudo iotop -o to find I/O-heavy processes in real time. A hungry database or backup job can skew your await numbers. Once you identify the culprit, think about optimizing the workload or moving it elsewhere.

Q: Should I enable TRIM on SSDs?

Absolutely—run sudo systemctl enable fstrim.timer to keep SSDs peppy. Without TRIM, writes slow down over time as the drive juggles garbage collection. It’s a quick win for SSD performance.

Quick Fix Summary:

To check disk await in SAR output, run sar -d in the terminal. Look for the await column in the output. If await is consistently above 50ms, dig into disk performance—check queue depth and disk health.

What does “await” mean in SAR output?

Await measures the average time, in milliseconds, for I/O requests to finish on a storage device.

That includes both the time requests wait in the queue and the time the device spends handling them. For example, an await of 30ms means I/O requests take about 30ms to complete on average. When await climbs above 100ms, you’ve likely hit a disk bottleneck—especially on systems pushing heavy I/O loads.

How do I check disk await with SAR?

Run sar -d in a terminal to see disk I/O stats and locate the await column.

Start by opening a terminal on your Linux box. Then fire off:

sar -d 1 3

The -d flag pulls disk activity reports. The 1 sets a one-second reporting interval, and 3 caps the output at three reports.

Scan the output for the await column. Here’s a quick sample:

DEV	tps	rd_sec/s	wr_sec/s	await	svctm	%util
sda	120.5	800	400	45.2	2.1	25.3

What should I do if await is high?

If await is consistently above 50ms, inspect disk health and queue depth right away.

Start with disk health:

Run smartctl -a /dev/sda (you’ll need smartmontools installed).
Check queue depth with iostat -x 1 and watch for high avgqu-sz numbers.

How can I use iostat to dig deeper?

Run iostat -x 1 3 to pull extended disk stats, including await, %util, and avgqu-sz.

This gives you a clearer picture of whether the slowdown is disk-bound or queue-driven. Picture a %util near 100% alongside a sky-high await—that screams a saturated disk.

Where do I look for disk errors?

Check dmesg | grep -i error or /var/log/messages for disk-related errors.

Failing disks often push await through the roof. If you spot errors, swap the drive immediately—hardware issues won’t fix themselves.

Can a single process cause high await?

Yes—use sudo iotop -o to find I/O-heavy processes in real time.

A hungry database or backup job can skew your await numbers. Once you identify the culprit, think about optimizing the workload or moving it elsewhere.

How often should I monitor await?

Monitor disk performance regularly—set alerts for await above 50ms.

Tools like sar, iostat, or Prometheus + Grafana are perfect for tracking trends. Build a Grafana dashboard that pings you the moment await spikes.

What RAID setup lowers await best?

RAID 10 generally delivers the best I/O balance for most workloads.

If you’re stuck on HDDs, RAID 10 cuts latency better than RAID 5 or 6. For SSDs, the gains shrink, but redundancy still matters.

Should I enable TRIM on SSDs?

Absolutely—run sudo systemctl enable fstrim.timer to keep SSDs peppy.

Without TRIM, writes slow down over time as the drive juggles garbage collection. It’s a quick win for SSD performance.

Does filesystem choice affect await?

Yes—XFS often edges out ext4 for high-throughput workloads.

In most cases, XFS handles large files and heavy I/O better than ext4. If you’re chasing lower await, give XFS a try.

How do I balance I/O across disks?

Spread the load—use LVM striping or move data/logs to separate disks.

Don’t let one disk choke under pressure. Stripe volumes with LVM or split database files and logs across drives. Honestly, this is the best way to keep await in check.

When should I upgrade from HDD to SSD?

Upgrade when HDDs routinely show await above 50ms under load.

SSDs usually land under 10ms, while HDDs can easily hit 50ms or worse. For anything mission-critical, NVMe SSDs shave off even more latency.