Picking Alert Profiles for SaaS and VOIP Targets

Picking Alert Profiles for SaaS and VOIP Targets

Remote users depend heavily on cloud and SaaS applications for daily productivity and real-time collaboration.
Detecting outages and performance degradation early requires alerting that is application-aware and tuned for remote network conditions.

This article provides recommended alert profiles for common SaaS services using four core metrics:

  • Network Latency
  • Packet Loss
  • DNS Resolution Time
  • HTTP Response Time

Alerting is based on three supported profile types:

:white_check_mark: Up/Down Alerts
:white_check_mark: Performance Baseline Alerts (short-term vs long-term average)
:white_check_mark: Performance Watermark Alerts (short-term vs fixed threshold)


Alert Profile Types (Supported)

:white_check_mark: 1. Up/Down Alerts (Outage Detection)

Triggered when a test fails completely for a sustained period.

Examples:

  • No ping reply
  • DNS lookup fails
  • HTTP request cannot complete

Recommended trigger: 2–3 minutes of continuous failure


:white_check_mark: 2. Performance Baseline Alerts (Degradation vs Normal)

Triggered when short-term performance deviates significantly from historical behavior.

Example rule:

  • Short-term latency average (5–10 min)
    is 1.5× worse than long-term average (7–30 days)

Best for detecting:

  • ISP congestion
  • Regional SaaS slowdown
  • Wi-Fi degradation
  • Routing changes

:white_check_mark: 3. Performance Watermark Alerts (Threshold-Based)

Triggered when a short-term metric exceeds a fixed limit.

Best for enforcing:

  • SLA expectations
  • Voice/video quality thresholds
  • Consistent org-wide standards


Recommended SaaS Alert Targets & Profiles

Below are practical alert thresholds and profiles for common SaaS applications.


Microsoft Teams

Target

  • teams.microsoft.com

Up/Down Alerts

  • Ping failure for 2–3 minutes
  • DNS resolution failure for 2 minutes

Baseline Alerts

  • Latency short-term > 1.5× baseline
  • Packet loss short-term > 2× baseline

Watermark Alerts

Metric Alert Threshold
Latency > 150 ms
Packet Loss > 2%
DNS Time > 200 ms
HTTP Response > 2 sec

Teams is highly sensitive to packet loss.


Zoom

Target

  • zoom.us

Up/Down Alerts

  • Ping failure for 2 minutes
  • DNS failure for zoom.us

Baseline Alerts

  • Latency short-term > 1.5× baseline
  • Packet loss short-term > 2× baseline

Watermark Alerts

Metric Alert Threshold
Latency > 160 ms
Packet Loss > 2%
DNS Time > 250 ms
HTTP Response > 2 sec

Generic Cloud VoIP System

Target

  • Customer SBC / VoIP domain endpoint

Up/Down Alerts

  • No ping/SIP reachability for 1–2 minutes
  • DNS failure for VoIP host

Baseline Alerts

  • Latency short-term > 1.3× baseline
  • Packet loss short-term > 2× baseline

Watermark Alerts (Strict Voice Thresholds)

Metric Alert Threshold
Latency > 120 ms
Packet Loss > 1%
DNS Time > 150 ms

VoIP traffic is the most sensitive workload.


Generic Website

Target

  • Customer-defined domain

Up/Down Alerts

  • HTTP failure for 3 minutes
  • DNS resolution failure

Baseline Alerts

  • HTTP response > 2× baseline
  • DNS time > 1.5× baseline

Watermark Alerts

Metric Alert Threshold
HTTP Response > 3 sec
DNS Time > 300 ms
Latency > 200 ms
Packet Loss > 3%

Salesforce

Target

  • Tenant domain / salesforce.com

Up/Down Alerts

  • HTTP test failure for 3 minutes
  • DNS resolution failure

Baseline Alerts

  • HTTP response > 1.7× baseline
  • DNS time > 1.5× baseline

Watermark Alerts

Metric Alert Threshold
HTTP Response > 2.5 sec
DNS Time > 250 ms
Latency > 180 ms
Packet Loss > 2%

Salesforce issues are usually HTTP-related more than loss-related.


Google Meet

Target

  • meet.google.com

Up/Down Alerts

  • Ping failure for 2 minutes
  • DNS failure

Baseline Alerts

  • Latency > 1.5× baseline
  • Loss > 2× baseline

Watermark Alerts

Metric Alert Threshold
Latency > 150 ms
Packet Loss > 2%
DNS Time > 200 ms
HTTP Response > 2 sec

Gmail

Target

  • mail.google.com

Up/Down Alerts

  • HTTP failure for 3 minutes
  • DNS lookup failure

Baseline Alerts

  • HTTP response > 2× baseline

Watermark Alerts

Metric Alert Threshold
HTTP Response > 2.5 sec
DNS Time > 250 ms
Latency > 200 ms
Packet Loss > 3%

Google Drive

Target

  • drive.google.com

Up/Down Alerts

  • HTTP failure for 3 minutes
  • DNS lookup failure

Baseline Alerts

  • HTTP response > 2× baseline

Watermark Alerts

Metric Alert Threshold
HTTP Response > 3 sec
DNS Time > 250 ms
Latency > 180 ms
Packet Loss > 2%

File sync and uploads are impacted quickly by packet loss.


SharePoint

Target

  • Microsoft 365 tenant domain

Up/Down Alerts

  • HTTP failure for 3 minutes
  • DNS failure

Baseline Alerts

  • HTTP response > 1.8× baseline

Watermark Alerts

Metric Alert Threshold
HTTP Response > 3 sec
DNS Time > 250 ms
Latency > 180 ms
Packet Loss > 2%


Recommended Fixed Threshold Summary

Application Type Latency Packet Loss DNS HTTP
VoIP / Voice 120 ms 1% 150 ms
Teams / Zoom / Meet 150–160 ms 2% 200–250 ms 2 sec
Business SaaS (Salesforce) 180 ms 2% 250 ms 2.5 sec
Websites / Portals 200 ms 3% 300 ms 3 sec

Best Practices for Remote User Alerting

:white_check_mark: Use Up/Down alerts for outages
:white_check_mark: Use Baseline alerts to catch real-world degradation
:white_check_mark: Use Watermark alerts to enforce SLA-quality performance
:white_check_mark: Always tune baseline windows per user and geography


Suggested Averaging Windows

Alert Type Short-Term Window Long-Term Window
Baseline 5–10 minutes 7–30 days
Watermark 3–5 minutes Fixed

This alert profile structure is sound. In order to properly monitor SaaS + VoIP, **up/down, baseline, and watermark** should be separated.

A few brief additions:
* While packet loss is important for **VoIP**, jitter is as important. A 1% loss combined with a lot of jitter can ruin calls. Ping is insufficient for **Microsoft Teams / Zoom**. You may have a clean ICMP and yet have poor call quality since media operates via UDP. * For distant users, baseline notifications are much more helpful than static watermark levels. ISPs differ too much. * DNS resolution time is underestimated; it often peaks prior to complete SaaS deterioration.

One generic alert profile for everything is the greatest error I’ve seen. A simple website, Salesforce, and VoIP act differently. The distinction between signal and noise is the tuning according to workload.