[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC)：2025-09-04。"],[],[],null,["# Troubleshoot Performance Dashboard\n\nUse the following guidance to troubleshoot common issues with Performance Dashboard.\n\nShort spikes in packet loss that last a few minutes\n---------------------------------------------------\n\nCertain short spikes lasting for a few minutes are expected or possible.\nCheck for sustained loss that lasts for greater than five minutes. If you see\nsuch loss, or if the loss impacts your network or application performance, see\n[Getting support](/network-intelligence-center/docs/performance-dashboard/support/getting-support).\n\nSpikes in latency\n-----------------\n\nLatency is measured over Layer 4 traffic and is impacted by the TCP/IP stack\nrunning on the remote VM. Some applications with unique TCP/IP stack\nimplementations (such as VPN applications) might artificially increase latency\nin Performance Dashboard. If latency spikes despite the absence of\ncustom applications that might impact the TCP/IP stack, see\n[Getting support](/network-intelligence-center/docs/performance-dashboard/support/getting-support).\n\nNo data in the project Performance Dashboard\n--------------------------------------------\n\nIf you can't see packet loss information, make sure that you have virtual\nmachine (VM) instances in the Virtual Private Cloud (VPC) or Shared VPC\nnetwork that you are viewing. Performance Dashboard uses active probes\nto measure packet loss between zones. The number of active probes is related to\nthe number of VMs in the network. If the network that you are viewing has no\nVMs, there are no active probes and no packet loss measurement.\n\nIf you can't see latency information, make sure that you have VM instances\ngenerating enough traffic.\n\nPacket loss by zone heatmap has empty cells (marked as `N/A`)\n-------------------------------------------------------------\n\nThis is possible if a project is part of a Shared VPC. In such a case,\nthe matrix might include data for source zones where the project itself doesn't\nhave VMs, but other projects in the same Shared VPC do have VMs. Packet\nloss between two projects that use Shared VPC is measured toward the\ndestination service project.\n\nPacket loss seen in Performance Dashboard but not in application\n----------------------------------------------------------------\n\nPacket loss is based on active probing, which covers the entire mesh of VMs that\ncan communicate with each other. Your traffic pattern might only use a subset of\nthis mesh. It might be that your application is not utilizing the problematic\nnetwork path.\n\nAnother possibility is that not enough measurements are available to give high\nconfidence results. This might happen if there are very few VMs in a zone or if\nthe number of VMs fluctuates.\n\nPerformance Dashboard also includes preemptible VMs in the packet loss\ncalculation. Projects with a high percentage of such VMs might show a higher\npacket loss rate when VMs are getting preempted. Check if the higher packet loss\nrate corresponds to the time window when preemptible VMs are being shut down.\n\nConstant value seen for latency between two zones or between a region and an Internet location\n----------------------------------------------------------------------------------------------\n\nLatency data is stored as a distribution. This means that if all measurements\nconsistently fall in the same bucket in the distribution (which is more likely\nto happen with higher latency values), the resulting graph displays a constant\nvalue that is the mean of the bucket interval.\n\nFor example, if there is a bucket for every 10 ms\n(0 ms-10 ms, 10 ms-20 ms, 20 ms-30 ms, and so on), and there are three different\nlatency measurements of 251 ms, 256 ms, and 253 ms, all of them are in\nthe 250 ms-260 ms bucket. The latency is displayed as 255 ms, which is the mean\nof the 250 ms-260 ms bucket interval. If a new measurement of 259 ms is\nrecorded, the latency still displays as 255 ms.\n\nTroubleshooting Google Cloud performance metrics\n------------------------------------------------\n\n### Google Cloud performance appears more stable than project performance\n\nBecause the Google Cloud performance metrics are aggregated across all\nGoogle Cloud projects, they show fewer variations. This indicates that the typical\nperformance in a zone pair or a region, Internet location pair remains stable\neven though the performance of individual projects might show greater variation\ndue to randomness, natural variation, and measurement noise.\n\n### Can choose Google Cloud zones in the Google Cloud performance view but not in project view\n\nFor your project-specific view, you can only choose zones where the project is\ndeployed. The Google Cloud performance view enables you to view performance\nfrom any Google Cloud zone to any other Google Cloud zone.\n\nFor Internet to Google Cloud traffic type, you can only choose regions\nwhere there is traffic between VMs and Internet endpoints.\n\n### No project data in a zone pair but the Google Cloud performance data exists\n\nGoogle Cloud performance metrics are estimated by aggregating across all\nGoogle Cloud projects and are always available.\n\nNo project data in a (Google Cloud region, internet endpoint) pair but the Google Cloud performance data exists\n---------------------------------------------------------------------------------------------------------------\n\nGoogle Cloud performance metrics are estimated by aggregating across a\nwide range of Google Cloud to internet connections. The per project view\nshows only data for connections which were seen in your project with sufficient\ntraffic.\n\nWhat's next\n-----------\n\n- [Get support](/network-intelligence-center/docs/performance-dashboard/support/getting-support)"]]
