What's new in VictoriaMetrics * New releases ● v1.97 - long-term support (LTS) release ● v1.98 - mTLS authorization in all VictoriaMetrics components ● v1.99 - improved propagation of label filters in MetricsQL queries ● v1.100 - improvements in streaming aggregation and vmauth * New features ** DataDog integrations ● Accept data from new versions of DataDog agent via DataDog v2 API ● Accept data from DataDog AWS lambda extensions ** OpenTelemetry Data ingestion via Amazon Firehose: Accept OTEL metrics from Amazon Firehose Prometheus-compatible naming ● -opentelemetry.usePrometheusNaming command-line flag ● Converts metric names and labels into Prometheus-compatible naming in the same way as OTEL collector does ** Hetzner service discovery ● Discover scrape targets at Hetzner Cloud and Hetzner Robot ● hetzner_sd_configs ** Per-tenant and per-label downsampling (enterprise feature) ● Individual downsampling configs per distinct sets of tenants ● Individual downsampling configs per distinct sets of time series ** New Graphite query functions ● aggregateSeriesLists ● diffSeriesLists ● multiplySeriesLists ● sumSeriesLists ** mTLS support ● Every VictoriaMetrics Enterprise component verifies client TLS certificates if -mtls command-line flag is set ● Vmauth Enterprise supports mTLS-based request routing ** vmauth: improved request routing ● Host-based request routing ● Query args-based request routing ● Arbitrary header value-based request routing ** vmauth: DNS-based load balancing ● Even load balancing across IP addresses behind a single hostname ● Works great with vminsert and vmselect services MetricsQL: new functions ● sum_eq_over_time ● sum_gt_over_time ● sum_le_over_time ● count_values_over_time - counts the number of unique sample values over time ** MetricsQL: improved label filters’ propagation ** Stream aggregation improvements ● Reduced memory usage by up to 5x ● De-duplication during data ingestion ● New aggregation functions: ○ unique_samples ○ increase_prometheus ○ total_prometheus ● Ability to drop labels before stream aggregation and de-duplication ● keep_metric_names option ** Performance improvements ● Improve registration speed for new time series by up to 10x ● Reduce memory usage when scraping targets with big number of metrics ● Optimize performance for /api/v1/labels and /api/v1/label/.../values APIs