Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- Uptime is online slots divided by valid slots; coverage is valid slots divided by scheduled slots. Read both.
- PuppyIP aggregates regional test nodes into five-minute slots. Any recorded outage makes the slot offline; all normal makes it online; other cases are unknown.
- The 30-day window uses Asia/Shanghai calendar days: previous 29 days plus today. Today includes only reached, applicable scheduled times.
- Initialization 100% is not historical measurement. Missing data is not normal, and regional test monitoring guarantees neither every purchased IP nor target-platform access.
100% alone cannot establish long-term stability
100% uptime need not mean uninterrupted service for 30 days. First distinguish initialization from measurement, then check valid samples, scheduled samples and coverage. All valid observations can be online while many scheduled times lack evidence, giving 100% uptime with low coverage.
This explains regional test monitoring on PuppyIP's free-test page for pre-purchase reading. Examples are fictional, not current regional results. No resource was claimed or connection tested for a reader.
The monitored objects are regional test nodes
Results aggregate nodes that started monitoring in the selected country/region. Each regional five-minute time counts once, not once per node and then averaged; raw probe-request success is not automatically the uptime metric.
New nodes join only applicable times after monitoring begins; present normal status cannot fill earlier history. Objects are not every purchased IP, your local network, a target website or login success. Recheck dates and data after changing region.
Each slot can be online, offline or unknown
If any participating node records offline at a time, the regional slot is offline; other normal nodes do not cancel it. All expected participants must have normal records for online. Without outage evidence but with a missing/undetermined record, it is unknown.
Unknown means neither confirmed outage nor normal. One normal node plus another missing record cannot establish regional online status. Conversely, a missing node must not hide another's explicit outage.
Uptime and coverage use different denominators
Valid slots equal online plus offline. Uptime equals online divided by valid slots, times 100%. Unknown slots are excluded, not assumed successful or counted as outages without evidence.
Coverage equals valid divided by scheduled slots, times 100%. It measures evidence completeness, not how much was online. With zero valid slots uptime is unknown; if scheduled slots are positive coverage is 0%. If scheduled slots are also zero, coverage is unknown, not forced to 0% or 100%.
Fictional example: 100% uptime with 80% coverage
Suppose ten scheduled slots include eight online, zero offline and two unknown. This is fictional, not PuppyIP measurement. Eight valid slots give 8/8=100% uptime and 8/10=80% coverage. Say all eight valid slots were online with two undetermined slots, not that the entire window had no interruption.
If those eight become seven online and one offline, uptime becomes 87.5% while coverage stays 80%. Completing monitoring evidence and reducing outages are distinct; a more attractive percentage cannot replace the other.
Today is an unfinished Shanghai calendar day
Asia/Shanghai divides dates into the previous 29 calendar days plus today, not exactly 30 times 24 hours backward from now or node-local midnights. Crossing Shanghai midnight starts another calendar day.
A full day has 288 five-minute times, but times before monitoring starts do not automatically count as scheduled. Today includes only reached times with eligible nodes, not future times as missing. Its changing figures are not equally complete evidence to a finished historical day.
Do not average daily percentages
Overall uptime sums online and valid slots over 30 days before division; coverage sums valid and scheduled slots. Days with many versus few valid records have different weights. Averaging daily percentages loses denominators.
Cell colors and completeness differ. Green can mean all valid records online even with missing historical slots or less than a full day after start. It establishes no evidence for every scheduled time; read coverage and counts alongside color.
Initialization and rechecks do not rewrite history
Labels marking initialization display and not historical measurement distinguish display from data. Initialization 100% generates no observations. Actual records subsequently govern display rather than continuously filling missing days green. Retain the label when quoting, not an isolated 100%.
Fault rechecks confirm resource state separately. A slot's initial offline record does not become normal because a later recheck recovers; retries do not add scheduled slots. Present success cannot backfill checks never made during an interruption.
Preserve scope when citing or comparing
Confirm region, objects, date range, Shanghai time and update time; distinguish initialization/measurement and read uptime, coverage and valid/scheduled counts. Label today in progress. Unknown counts stay unknown rather than inferred as precise samples from percentages.
Compare equivalent objects, windows and definitions where possible. Without comparable evidence, explain gaps. Do not invent a qualifying coverage threshold or promise future performance from a screenshot.
Validate your client instead of substituting statistics
Regional history helps explain evidence but does not replace client protocol/authentication, exit IP, DNS, WebRTC and IPv6 checks. For a short trial, read eligibility and prepare the client first. Claims remain subject to public scope, personal eligibility and regional inventory; a limited trial cannot measure long-term stability.
See the trial guide and acceptance checklist in Sources. Monitoring percentages are not a product SLA, speed promise or target-platform guarantee. Use actual public/written service conditions rather than deriving them from the test page.
Sources
Frequently Asked Questions
Does 100% mean no outage for 30 days?
No. Exclude initialization and check valid/scheduled counts and coverage. Missing slots remain unknown, and today is unfinished.
How do uptime and coverage differ?
Uptime is online/valid, where valid is online+offline. Coverage is valid/scheduled. One measures determined results and the other completeness.
Is no data offline or 100%?
Neither by default. Uptime with zero valid slots is unknown. Coverage is 0% if scheduled is positive, unknown if also zero. Initialization 100% is display only.
Does one normal node make a region online?
No. Any outage makes it offline; all expected nodes normal makes it online. A missing node with no outage gives unknown, counted once per regional time.
Does a small count today mean a whole day missed?
No. Only reached applicable slots count. Future times are not scheduled yet, and Shanghai calendar today remains in progress.
Does the test page guarantee my account or purchased IP?
No. Regional test-node statistics are neither a full-product SLA nor target guarantee. Authorized actual-client acceptance and separate eligibility/service terms still apply.