Try each one before opening the answer. Every option is explained, with the Google Cloud documentation page that proves it.
Question 1 · choose 1
Users report intermittent failures when calling a service behind an internal passthrough Network Load Balancer with four backend VMs. A Connectivity Test from a client VM to the forwarding rule shows the configuration analysis as reachable, and the live data plane analysis as partially reachable, with 70% of the probes delivered. What does this result suggest?
AEvery backend delivers traffic, because partially reachable means at least one probe reached each of the four backends
BLatency on the path is too high, because partially reachable means the probes arrived above a delay threshold
CThe test is invalid, because live data plane analysis doesn't send probes to load balancer backends
DSome probes went to a backend that didn't deliver them, so the four backends should be checked one by one
Show the answer and why
AEvery backend delivers traffic, because partially reachable means at least one probe reached each of the four backends
Incorrect
Probes to a load balancer go to random backends, so some backends may get no probe at all; the status only counts the share of probes delivered.
BLatency on the path is too high, because partially reachable means the probes arrived above a delay threshold
Incorrect
The status depends on the share of probes delivered; latency is reported separately.
CThe test is invalid, because live data plane analysis doesn't send probes to load balancer backends
Incorrect
For a load balancer with several backends, each probe is sent to a random backend.
DSome probes went to a backend that didn't deliver them, so the four backends should be checked one by one
Correct
Fewer than 95% but more than none arrived, and each probe went to a random backend, which points at individual backends.
Connectivity Tests combines a configuration analysis with, where supported, a live data plane analysis that sends TCP or UDP probes. Reachable means at least 95% of probes arrived, unreachable means none, and anything between is partially reachable. With a load balancer, each probe goes to a random backend, so some backends may be tested by many probes and others by none; a partial result with a clean configuration points at individual backends.
A VPC network has these ingress allow rules: P900 allows TCP 22 from 10.0.0.0/8 to all instances; P1000 allows TCP 22 from 10.20.0.0/16 to instances tagged bastion. To save costs, Firewall Rules Logging (firewall policy rules logging) is disabled on every rule. Which Firewall Insights finding can still flag P1000 as redundant?
AAn allow rule with no hits insight for P1000
BAn allow rule with unused attributes insight for P1000
CA shadowed firewall rule insight for rule P1000
DA deny rule with no hits insight for P1000
Show the answer and why
AAn allow rule with no hits insight for P1000
Incorrect
Hit-based insights depend on firewall logging data, which isn't collected for these rules.
BAn allow rule with unused attributes insight for P1000
Incorrect
Unused attributes are found from observed traffic, which needs firewall logging data.
CA shadowed firewall rule insight for rule P1000
Correct
Shadowed rule insights come from the rule configuration: P900 has a higher priority and covers every connection P1000 allows.
DA deny rule with no hits insight for P1000
Incorrect
That insight covers deny rules over the observation period; P1000 is an allow rule.
Firewall Insights offers shadowed rule insights, derived from how the rules are configured, and overly permissive insights such as allow rules with no hits or with unused attributes, which rely on usage data from firewall logging. A rule whose matches are fully covered by a rule of higher or equal priority is shadowed, with or without logging.
A Connectivity Test from a VM to a partner's public IP address on the internet includes a live data plane analysis. How far does that analysis test the path, and how are its results shown?
AAll the way to the partner's server, with one result for the whole path
BNot at all: live data plane analysis never runs for internet destinations
COnly within the VM's subnet, because probes stop at the first router
DFrom the VM to Google's network edge, with results for each edge router
Show the answer and why
AAll the way to the partner's server, with one result for the whole path
Incorrect
When the destination isn't a Google Cloud resource, probes test only up to the network edge.
BNot at all: live data plane analysis never runs for internet destinations
Incorrect
It runs for internet destinations, up to Google's network edge.
COnly within the VM's subnet, because probes stop at the first router
Incorrect
Probes travel from the source endpoint to the network edge location.
DFrom the VM to Google's network edge, with results for each edge router
Correct
For an internet destination, probes test connectivity to the network edge and results are shown per Google edge router.
Live data plane analysis sends TCP or UDP probes and reports delivered probes, a reachability status and latency. If the destination isn't a Google Cloud resource, connectivity is tested between the source and the network edge location, and for an internet address the results are shown for each Google network edge router that traffic might go through.
A team opens the project performance view of Performance Dashboard and finds packet loss data for us-central1 but none for europe-west4, where it plans to deploy next month. What explains the missing data?
Aeurope-west4 doesn't support Performance Dashboard
BThe project view shows metrics only for zones where the project has VM instances
CVPC Flow Logs must first be enabled on a subnet in europe-west4
DPacket loss data appears only after a Connectivity Test is run there
Show the answer and why
Aeurope-west4 doesn't support Performance Dashboard
Incorrect
The project view's coverage depends on where the project has VMs, not on region support.
BThe project view shows metrics only for zones where the project has VM instances
Correct
In the project performance view, packet loss and latency appear only for zones with the project's VMs.
CVPC Flow Logs must first be enabled on a subnet in europe-west4
Incorrect
Performance Dashboard doesn't depend on VPC Flow Logs being enabled.
DPacket loss data appears only after a Connectivity Test is run there
Incorrect
Performance Dashboard doesn't need Connectivity Tests to show data.
Performance Dashboard has a project performance view and a Google-wide view. The project view shows packet loss or latency only for zones where the project has VM instances, so a region without VMs has no project data; the Google-wide view covers Google Cloud's network as a whole.
A network engineer wants to explore which VMs and destinations exchange the most traffic, based on VPC Flow Logs, without writing complex SQL queries against exported logs. Which tool fits?
ANetwork Analyzer
BFirewall Insights
CFlow Analyzer
DConnectivity Tests
Show the answer and why
ANetwork Analyzer
Incorrect
Network Analyzer reports configuration problems and utilization warnings, not flow-by-flow traffic exploration.
BFirewall Insights
Incorrect
Firewall Insights covers firewall rule usage.
CFlow Analyzer
Correct
Flow Analyzer lets you understand VPC traffic flows without writing complex SQL queries for VPC Flow Logs.
DConnectivity Tests
Incorrect
Connectivity Tests analyzes whether a path works, not observed traffic volumes.
Flow Analyzer, part of Network Intelligence Center, analyzes VPC Flow Logs so that you can quickly understand VPC traffic flows without writing complex SQL queries, with SQL filters available to refine results.