Hosted Graphite — Complete Visibility for Heroku Connect Teams

LIVE DEMO DASHBOARDS

YOUR HEROKU CONNECT STACK,
VISIBLE TO YOUR WHOLE TEAM

Auto-provisioned dashboards for every service your team runs. Shared across everyone — from the junior dev to the on-call SRE — the moment you install.

WEB.1 — Memory Usage

341 MB

NORMAL

Response Time P95

142 ms

HEALTHY

WEB.1 — Load Avg23%

WEB.2 — Load Avg31%

WORKER.1 — Load Avg42%

14:32:01heroku.routerat=info method=GET path="/api/v1/orders" service=98ms status=200

14:32:04heroku.web.1source=web.1 dyno=heroku.1 sample#memory_total=341.2MB

14:32:07heroku.routerat=info method=POST path="/api/v1/sync" service=203ms status=201

14:32:09heroku.web.2source=web.2 sample#load_avg_1m=0.31

14:32:12alert.engine✓ All dyno metrics within thresholds

14:32:15heroku.routerat=info path="/health" service=8ms status=200

Active Connections

17 / 25

NORMAL

Cache Hit Ratio

99.1 %

EXCELLENT

Table Bloat4.2%

Index Efficiency97.8%

Disk Usage61%

14:33:00postgres.dbsample#db_size=3.2GB sample#tables=47 sample#active-connections=17

14:33:05postgres.dbsample#cache-hit-rate=0.991 sample#index-cache-hit-rate=0.998

14:33:10alert.engine⚠ Disk usage at 61% — approaching 70% threshold

14:33:10notify.slackSent alert to #heroku-ops: Disk usage trending up

Memory Usage

78 %

WARNING

Hit Rate

94.3 %

GOOD

Cache hits vs misses94.3%

⚠ ALERT FIRING — redis.memory_used_ratio

Condition: value > 0.75 for 5 minutes | Current: 0.78 | Threshold: 0.75
Channels: #heroku-ops (Slack), [email protected]

14:34:01redis.hksample#memory_used_ratio=0.78 sample#hit_rate=0.943 sample#connected_clients=12

14:34:01alert.engine🔔 ALERT: redis.memory_used_ratio exceeded 0.75 threshold

14:34:02notify.slackPosted to #heroku-ops: Redis memory warning

Consumer Lag

0 msgs

CAUGHT UP

Throughput

3,241 msg/s

NORMAL

Partition utilization44%

orders-topic — consumer lag0 msgs

events-topic — consumer lag0 msgs

Partition count8 / 16

14:35:00kafka.hksample#consumer_lag=0 sample#throughput_msgs_per_sec=3241

14:35:00kafka.hksample#partition_count=8 sample#leader_count=8

14:35:02alert.engine✓ All Kafka consumer groups caught up

SALESFORCE ORG

production-org

Objects synced:12

API calls used:14,230 / 100k

⟷

HEROKU POSTGRES

DATABASE_URL

Rows synced:1.2M

Last sync:3s ago

→ Your team can overlay Heroku Connect sync windows against Postgres connection load — instantly visible to every engineer, not just the one who knows where to look:

DB Connections during sync window21/25

DB Connections outside sync11/25

Sync-triggered P95 latency spike+340ms

14:36:00hk-connectBulk sync started: Account, Contact, Opportunity (CDC)

14:36:01postgres.db⚠ active-connections spike: 11 → 21 (sync window)

14:36:03heroku.routerservice time P95 spike: 142ms → 482ms (connection contention)

14:36:12hk-connectBulk sync complete. 4,821 rows written to Postgres

Orders / Minute

247 / min

NORMAL

Payment Error Rate

0.3 %

LOW

SEND CUSTOM METRICS FROM YOUR APP CODE →

# Python example
import socket, os
def send_metric(name, value):
key = os.environ['HOSTEDGRAPHITE_APIKEY']
msg = f"{key}.{name} {value}\n"
sock = socket.socket()
sock.connect(('carbon.hostedgraphite.com', 2003))
sock.send(msg.encode())
send_metric('orders.per_min', 247)  # done ✓

BUILT FOR HEROKU CONNECT TEAMS

ONE PLATFORM.
EVERY ROLE ON YOUR TEAM.

Heroku Connect teams aren't just engineers — they're developers debugging sync lag, ops engineers watching dyno health, and leads who need the full picture when something goes wrong at 2am.

For Developers

STOP DEBUGGING IN THE DARK

When your Heroku Connect sync window coincides with a performance dip, you need to know immediately — not after trawling through log drains. See every layer of your stack in one dashboard.

Correlate Heroku Connect CDC sync windows with Postgres connection spikes
Pinpoint slow endpoints with per-path router metrics
Track custom app metrics alongside infra signals in the same view
Send metrics from any language — Python, Node.js, Ruby, Go, Java, PHP

For DevOps & SREs

ALERTS BEFORE YOUR USERS NOTICE

Redis memory creeping toward saturation. Postgres connections climbing during a Salesforce bulk write. Dyno R14 errors appearing under load. Know about it before it becomes an outage.

Threshold alerts on any metric — Slack, PagerDuty, OpsGenie, Teams, Email
Auto-provisioned dashboards for Dynos, Postgres, Redis, and Kafka
30-second metric resolution with up to 24 months retention
AWS CloudWatch integration for multi-cloud Heroku setups

For Engineering Leads

YOUR WHOLE TEAM, ONE PANE OF GLASS

Shared Grafana dashboards mean your entire team — junior devs, senior engineers, and stakeholders — sees the same data. No more "works on my machine" when production is struggling.

SAML SSO via your existing identity provider (Okta, Azure AD, etc.)
Team access controls — share dashboards across the whole org
Access Keys for read-only sharing with external stakeholders
Daily backups so your historical metric data is never at risk

For Heroku Connect Specifically

THE VISIBILITY CONNECT DOESN'T GIVE YOU

Heroku Connect tells you whether your Salesforce data synced. Hosted Graphite tells you what that sync did to your infrastructure — and whether your app can handle the next one.

Overlay sync timing against Postgres active-connection counts
Detect latency spikes caused by Bulk API write-back operations
Monitor Kafka topics used by event-driven Heroku Connect architectures
Track Heroku Postgres disk growth as your synced Salesforce data scales

TRANSPARENT PRICING

PREDICTABLE. SCALABLE.
NO SURPRISES.

Every plan includes dashboards, alerting, team access, and support from day one. Pay for metrics, not features — and scale only when you need to.

STARTER

^$19

/ month max

500 Metrics

Perfect for small Heroku apps — covers Dynos + Postgres + Redis automatically.

Automatic Heroku dashboards
6 months data retention
30s resolution
Alerting (Email, Slack)
Team access included
Hosted StatsD

⬡ Install on Heroku → Free Trial →

GROWTH

^$49

/ month max

1,000 Metrics

For teams running multiple services — Dynos, Postgres, Redis, Kafka + custom app metrics.

Everything in Starter
Kafka dashboards
Custom metric support
AWS CloudWatch integration
PagerDuty / OpsGenie alerts
Access Keys for sharing

⬡ Install on Heroku → Book Demo →

SCALE

^$149

/ month max

2,000 Metrics

For larger Heroku estates with many dynos, multi-app monitoring, and full Salesforce sync correlation.

Everything in Growth
12 months data retention
SAML SSO
Heroku Connect sync overlays
Multi-region support
Priority support

⬡ Install on Heroku → Book Demo →

Plans scale to 500,000 metrics. Enterprise pricing available. No contracts required.

READY TO GET STARTED?

GIVE YOUR WHOLE TEAM
HEROKU VISIBILITY.

Dynos. Postgres. Redis. Kafka. Heroku Connect. Custom metrics.
Shared dashboards, real-time alerts, and team access — from $19/month.

⬡ HEROKU ADD-ON

Install directly from the Heroku Elements Marketplace. Automatic dashboards for Dynos, Postgres, Redis, and Kafka appear the moment metrics start flowing. From $19/month, no contracts.

elements.heroku.com/addons/hostedgraphite →

⚡ FREE TRIAL

14 days completely free. No credit card required. Sign up at hostedgraphite.com and get instant access to dashboards, alerting, team features, and all integrations.

hostedgraphite.com/accounts/signup →

📅 BOOK A DEMO

30 minutes with a real engineer. We'll walk through your Heroku Connect setup, configure dashboards for your specific stack, and show you how your whole team can stay informed.

calendly.com/metricfire-sales/demo →

Trusted by Tripadvisor, Zynga, Dropbox · All Heroku regions · Starting at $19/month · No contracts

Heroku Connect · Metric Type 01 of 04

SYNC EVENTS

Every Heroku Connect sync cycle emits log events when it starts and completes. Hosted Graphite captures these as time-series metrics, letting your whole team see sync frequency, duration trends, and whether the last cycle finished successfully — without digging through raw log drains.

Install on Heroku → Book Demo

What are Sync Events?

Heroku Connect emits a log line each time a sync cycle begins and each time it completes. These events tell you that the sync engine is actively running, that it finished without hanging, and how frequently your Salesforce objects are being polled. With Accelerated Polling via CDC enabled, these events fire far more frequently for high-priority objects.

What Hosted Graphite captures

heroku.connect.sync.started — counter incremented each time a sync cycle begins
heroku.connect.sync.completed — counter incremented on successful completion
heroku.connect.sync.in_progress — gauge indicating whether a sync is currently active (1 or 0)
heroku.connect.sync.duration_ms — duration of each completed cycle in milliseconds

Example Log Lines (raw from Heroku)

# Sync started
app=my-app addon=heroku-connect at=info
  message="Sync started" object=Account
  timestamp=2026-03-25T14:32:00Z

# Sync completed
app=my-app addon=heroku-connect at=info
  message="Sync completed" object=Account
  rows_processed=1204 duration_ms=3821
  timestamp=2026-03-25T14:32:03Z
            

Recommended Alert Rules

⚠ ALERT — Sync stopped firing

Condition: heroku.connect.sync.started rate = 0 for > 15 minutes
Meaning: Connect may have stalled, lost its Salesforce OAuth token, or hit an API limit

⚠ ALERT — Sync never completing

Condition: started count > completed count by more than 3 over 10 minutes
Meaning: Syncs are starting but hanging — possible Postgres lock contention or Bulk API timeout

LIVE LOG STREAM — sync events LIVE

14:32:00heroku-connectSync started — object=Account

14:32:01heroku-connectSync started — object=Contact

14:32:03heroku-connectSync completed — Account rows=1204 duration=3821ms

14:32:05heroku-connectSync completed — Contact rows=892 duration=2104ms

14:32:06hosted-graphiteMetrics ingested: sync.started +2, sync.completed +2

14:32:30heroku-connectSync started — object=Opportunity (CDC)

OTHER METRIC TYPES

Heroku Connect · Metric Type 02 of 04

ROW COUNTS

Heroku Connect reports how many records were synced in each cycle per Salesforce object. Tracking this over time reveals data growth patterns, detects unexpected bulk operations, and lets your team monitor against row-limit thresholds before they become a billing or performance issue.

Install on Heroku → Book Demo

What are Row Count Metrics?

Each sync completion log includes a rows_processed field. Hosted Graphite extracts this per-object and builds time-series gauges showing how many records moved in each cycle. Over time, these charts reveal growth trends, abnormal spikes from bulk Salesforce operations, and how close you are to your Connect row quota.

What Hosted Graphite captures

heroku.connect.rows.processed — rows synced per object per cycle (gauge)
heroku.connect.rows.total — cumulative total rows tracked in your Connect instance
heroku.connect.rows.written_to_sf — rows written back to Salesforce from Postgres (write-direction)
heroku.connect.rows.written_to_pg — rows written from Salesforce into Postgres (read-direction)

Example Log Lines (raw from Heroku)

# Per-object row count on completion
app=my-app addon=heroku-connect
  message="Sync completed"
  object=Opportunity
  rows_processed=4821
  rows_inserted=312
  rows_updated=4509
  rows_deleted=0
  direction=sf_to_pg

# Totals in periodic status log
total_synced_rows=1204892
row_limit=5000000
utilisation_pct=24.1
            

Recommended Alert Rules

⚠ ALERT — Unexpected bulk operation

Condition: rows_processed for any single object > 10,000 in one cycle
Meaning: Someone ran a bulk update in Salesforce that could spike Postgres write load

⚠ ALERT — Row quota approaching

Condition: heroku.connect.rows.total > 80% of contracted row limit
Meaning: Time to review your Connect plan tier or archive old Salesforce records

LIVE LOG STREAM — row counts LIVE

14:33:01heroku-connectAccount sync complete — rows=1204 ins=12 upd=1192 del=0

14:33:03heroku-connectContact sync complete — rows=892 ins=4 upd=888 del=0

14:33:05heroku-connectOpportunity sync complete — rows=4821 ins=312 upd=4509 del=0

14:33:05hosted-graphite⚠ rows.processed spike: Opportunity 4821 (threshold: 2000)

14:33:05alert.engineSlack alert sent: #heroku-ops — bulk Opportunity write detected

OTHER METRIC TYPES

Heroku Connect · Metric Type 03 of 04

ERRORS

Heroku Connect logs sync failures, write-back errors, and conflict events. These are the signals your team most needs to see immediately — a write error left unresolved blocks the entire queue for that object. Hosted Graphite surfaces error rates as metrics with alerting so your team responds before data integrity degrades.

Install on Heroku → Book Demo

What are Error Metrics?

When Heroku Connect fails to write a record back to Salesforce, or when the Salesforce API returns a validation error, Heroku Connect logs it and pauses the queue for that object until the error is manually resolved. Tracking error rate as a metric — not just a log event — means your team gets alerted the moment errors begin accumulating, before the queue backs up.

What Hosted Graphite captures

heroku.connect.errors.total — cumulative error count across all objects
heroku.connect.errors.write_failed — failed writes to Salesforce (Ordered Writes queue blocking)
heroku.connect.errors.conflict — records where both Salesforce and Postgres were modified simultaneously
heroku.connect.errors.api_limit — errors caused by Salesforce API limit exhaustion
heroku.connect.errors.rate — errors per minute, derived metric for alerting

Example Log Lines (raw from Heroku)

# Write error — blocks the queue
app=my-app addon=heroku-connect at=error
  message="Write failed" object=Contact
  sf_id=0031k000009xyzAAA
  error_code=FIELD_CUSTOM_VALIDATION_EXCEPTION
  error_msg="Email domain not permitted"

# Conflict — simultaneous edit in both systems
message="Conflict detected" object=Account
  sf_id=0011k000008abcAAA
  resolution="salesforce_wins"

# API limit hit
message="API limit reached"
  daily_api_calls_used=149823
  daily_api_calls_limit=150000
            

Recommended Alert Rules

🔴 CRITICAL — Write queue blocked

Condition: heroku.connect.errors.write_failed > 0 sustained for 5 minutes
Meaning: An unresolved INSERT error is blocking all subsequent writes for that object — data will drift

⚠ ALERT — Error rate spike

Condition: heroku.connect.errors.rate > 5 errors/min
Meaning: Systematic validation failures — likely a schema change in Salesforce or an app deploying bad data

⚠ ALERT — API limit approaching

Condition: daily_api_calls_used > 80% of daily_api_calls_limit
Meaning: Sync will throttle or fail if the limit is reached before midnight Salesforce reset

LIVE LOG STREAM — errors LIVE

14:34:01heroku-connectContact sync OK — 0 errors

14:34:04heroku-connectWrite FAILED — Contact 0031k...AAA FIELD_CUSTOM_VALIDATION_EXCEPTION

14:34:04heroku-connectQueue BLOCKED — Contact writes halted until error resolved

14:34:05hosted-graphiteerrors.write_failed incremented — Contact

14:34:05alert.engine🔴 CRITICAL alert fired → PagerDuty: Write queue blocked

OTHER METRIC TYPES

Heroku Connect · Metric Type 04 of 04

LATENCY

Heroku Connect's sync latency tells your team how fresh your Salesforce data actually is in Postgres — and how long each sync cycle takes. With CDC Accelerated Polling, latency drops dramatically for critical objects. Hosted Graphite makes this visible as a trend chart so you can see degradation before users report stale data.

Install on Heroku → Book Demo

What is Sync Latency?

Sync latency is the time between a record changing in Salesforce and that change appearing in your Heroku Postgres database. With standard polling it's typically in the range of 5–20 minutes. With CDC Accelerated Polling enabled for an object, it drops to seconds. Tracking this as a metric lets your team alert when latency degrades — e.g. due to Salesforce API slowdowns, large batch writes, or CDC channel contention.

What Hosted Graphite captures

heroku.connect.latency.sync_duration_ms — wall-clock time of each sync cycle per object
heroku.connect.latency.poll_interval_s — actual observed gap between poll cycles (vs configured)
heroku.connect.latency.data_age_s — estimated staleness of Postgres data for each object
heroku.connect.latency.api_response_ms — Salesforce API response time as observed by Connect
heroku.connect.latency.cdc_lag_ms — CDC event delivery lag for Accelerated Polling objects

Example Log Lines (raw from Heroku)

# Duration logged on completion
app=my-app addon=heroku-connect
  message="Sync completed" object=Opportunity
  duration_ms=3821
  poll_mode=cdc_accelerated
  cdc_event_lag_ms=142
  sf_api_time_ms=890

# Standard polling — longer interval
message="Sync completed" object=Product2
  duration_ms=12340
  poll_mode=standard
  poll_interval_s=600
  data_age_s=634
            

Recommended Alert Rules

⚠ ALERT — Sync duration spiking

Condition: sync_duration_ms P95 > 15,000 for any object over 10 minutes
Meaning: Salesforce API is slow, or a large batch is causing individual sync cycles to time out

⚠ ALERT — CDC lag elevated

Condition: cdc_lag_ms > 2,000 sustained for 5 minutes
Meaning: CDC event delivery is degraded — data freshness for accelerated objects is compromised

⚠ ALERT — Data age exceeding SLA

Condition: data_age_s > your_sla_threshold (e.g. 1200s = 20 min)
Meaning: Records in Postgres are stale — users may be seeing outdated Salesforce data in your app

LIVE LOG STREAM — latency LIVE

14:35:00heroku-connectOpportunity sync (CDC) — duration=3821ms cdc_lag=142ms ✓

14:35:00heroku-connectContact sync (CDC) — duration=2104ms cdc_lag=98ms ✓

14:35:00heroku-connectAccount sync (CDC) — duration=2980ms cdc_lag=110ms ✓

14:35:05heroku-connectProduct2 sync (standard) — duration=12340ms data_age=634s

14:35:05hosted-graphitelatency metrics updated — 5 objects tracked

OTHER METRIC TYPES

Heroku Connect keeps data in sync. We keep your team informed.

YOUR HEROKU CONNECT STACK,
VISIBLE TO YOUR WHOLE TEAM

ONE PLATFORM.
EVERY ROLE ON YOUR TEAM.

STOP DEBUGGING IN THE DARK

ALERTS BEFORE YOUR USERS NOTICE

YOUR WHOLE TEAM, ONE PANE OF GLASS

THE VISIBILITY CONNECT DOESN'T GIVE YOU

EVERY SERVICE YOUR
HEROKU CONNECT TEAM RUNS

DYNOS

POSTGRES

REDIS

KAFKA

HEROKU CONNECT

CUSTOM METRICS

PREDICTABLE. SCALABLE.
NO SURPRISES.

GIVE YOUR WHOLE TEAM
HEROKU VISIBILITY.

What are Sync Events?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What are Row Count Metrics?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What are Error Metrics?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What is Sync Latency?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

Heroku Connect keeps data in sync. We keep your team informed.

YOUR HEROKU CONNECT STACK,VISIBLE TO YOUR WHOLE TEAM

ONE PLATFORM.EVERY ROLE ON YOUR TEAM.

STOP DEBUGGING IN THE DARK

ALERTS BEFORE YOUR USERS NOTICE

YOUR WHOLE TEAM, ONE PANE OF GLASS

THE VISIBILITY CONNECT DOESN'T GIVE YOU

EVERY SERVICE YOURHEROKU CONNECT TEAM RUNS

DYNOS

POSTGRES

REDIS

KAFKA

HEROKU CONNECT

CUSTOM METRICS

PREDICTABLE. SCALABLE.NO SURPRISES.

GIVE YOUR WHOLE TEAMHEROKU VISIBILITY.

What are Sync Events?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What are Row Count Metrics?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What are Error Metrics?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

What is Sync Latency?

What Hosted Graphite captures

Example Log Lines (raw from Heroku)

Recommended Alert Rules

YOUR HEROKU CONNECT STACK,
VISIBLE TO YOUR WHOLE TEAM

ONE PLATFORM.
EVERY ROLE ON YOUR TEAM.

EVERY SERVICE YOUR
HEROKU CONNECT TEAM RUNS

PREDICTABLE. SCALABLE.
NO SURPRISES.

GIVE YOUR WHOLE TEAM
HEROKU VISIBILITY.