Getting topology data
1. Getting topology data
So far you have learned about various different topologies. If you are wondering if there is a way to visualize the topology of the network, the answer is yes. Network Topology does exactly that. Let’s learn about getting topology data in this section. Network Topology is a visualization tool that shows the topology of the network infrastructure. It serves as a map, illustrating the complex relationships between various components and how they interact. Network Topology offers an "infrastructure view" showcasing elements like VPC networks, hybrid connectivity, and connections to Google-managed services. It goes beyond static configurations by incorporating real-time operational data, revealing traffic paths, throughput, and the current state of connections between various workloads. By presenting this information in a graph format, where nodes and lines represent entities and connections, Network Topology simplifies the understanding of complex networking relationships. This visual representation aids in identifying bottlenecks, optimizing traffic flows, and troubleshooting network issues more effectively. Network Topology actively gathers real-time telemetry and configuration data from Google's infrastructure to create a visual representation of your network resources. It captures various elements like configuration details, performance metrics, and logs to deduce the relationships between resources within a single project or across multiple projects. By consolidating this information, Network Topology generates a comprehensive graph that accurately depicts your entire network deployment. Network Topology maintains a historical record spanning six weeks, offering valuable insights into past network configurations and interactions. This history is organized into hourly snapshots, commencing at the beginning of each hour. Each snapshot captures the fundamental entities and their communication patterns within that specific hour. Even if resources, like instances, are created and deleted within an hour, they are still documented in the snapshot for that duration. This ensures a comprehensive historical representation of your network's evolution. At the lowest level of each hierarchy are base entities, which represent individual resources capable of direct network communication, such as VM instances or GKE pods. When dealing with complex networks containing numerous base entities, a flat view can be overwhelming. To address this, Network Topology aggregates base entities into higher-level hierarchical entities that can be expanded or collapsed. Initially, the graph displays all base entities aggregated into their top-level hierarchy. For example, VM instances are aggregated into instance groups, then into zones. Meanwhile, GKE pods are aggregated into workloads, then namespaces, and finally into clusters. Each entity, whether base or hierarchical, is represented as a circular node in the graph, and each base entity has its own unique hierarchy. For example, load balancers are organized differently than VM instances within the graph. To simplify the view, Network Topology aggregates related resources into hierarchical entities, with each resource type having its own distinct hierarchy. Network Topology visually represents traffic between entities as lines, connecting entities if at least one side is sending traffic. It displays connections across various hierarchy levels, as long as the base entities communicate. For instance, a connection between two regions is shown if at least one VM instance in each region is communicating. The tool supports various traffic protocols (TCP, UDP, ICMP, ICMPV6, ESP, GRE) for specific paths, visualizing traffic within and across VPC networks, between Google Cloud and the internet, and to/from VPN gateways, Interconnect connections, and router appliances. In the GKE view, Network Topology shows traffic within a cluster (between pods on different nodes), and between pods on the same node if intranode visibility is enabled, and between clusters and external IPs. Let’s explore a use case. Charlie, a network administrator, is responsible for a network that includes several load-balanced applications. Charlie has been alerted to a latency issue causing the organization's mobile application to intermittently slow down and time out. Multiple users are experiencing this problem, and Charlie has confirmed that no recent application deployments have occurred. The issue is likely due to a change in the network environment rather than a problem with the application itself. Charlie needs to use the Network Topology tool to get a comprehensive visibility into the cloud network infrastructure. They do this because the graphical representation of the network topologies enables the engineers to quickly understand the connections between resources, identify potential bottlenecks, and to troubleshoot network issues efficiently. Beginning with the external clients in the Americas, Charlie selects the traffic metrics between that region and the shopping-site-lb load balancer. Network Topology then displays charts in the details pane, providing information such as ingress and egress traffic between the selected entities, along with key metrics like queries per second (QPS) and HTTP request latency. Upon reviewing the request latency chart, Charlie notices that the 50th, 95th, and 99th percentile values are all higher than expected. To gain a broader perspective on this latency issue, Charlie extends the time series charts to cover the past 6 weeks for further analysis. After expanding the timeframe, Charlie observes a significant jump in latency that occurred approximately two hours ago, coinciding with the initial reports of the issue. This confirms their suspicion that the increased latency is related to the load balancer. To investigate further, Charlie navigates to the Load balancing page in the Google Cloud console. Their analysis reveals that one of the instances in the load balancer's backend service was taking longer than usual to respond. By removing this problematic instance from service, Charlie successfully resolved the latency issue and restored normal application performance.2. Let's practice!
Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.