according to the provided Prometheus-compatible [scrape configs](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config)
and send data to multiple remote storage systems, vmagent has the following additional features:
- vmagent usually requires lower amounts of CPU, RAM and disk IO comparing to Prometheus when scraping big number of targets (more than 1000)
or targets with big number of exposed metrics.
- vmagent provides independent disk-backed buffers per each configured remote storage (aka `-remoteWrite.url`). This means that slow or temporarily unavailable storage
doesn't prevent from sending data to healthy storage in parallel. Prometheus uses a single shared buffer for all the configured remote storage systems (aka `remote_write->url`)
with the hardcoded retention of 2 hours.
- vmagent may accept, relabel and filter data obtained via multiple data ingestion protocols additionally to data scraped from Prometheus targets.
I.e. it supports both `pull` and `push` protocols for data ingestion.
### How does VictoriaMetrics compare to other remote storage solutions for Prometheus such as [M3 from Uber](https://eng.uber.com/m3/), [Thanos](https://github.com/thanos-io/thanos), [Cortex](https://github.com/cortexproject/cortex), etc.?
VictoriaMetrics is simpler, faster, more cost-effective and it provides [MetricsQL query language](MetricsQL) based on PromQL. The simplicity is twofold:
- It is simpler to configure and operate. There is no need in configuring [sidecars](https://github.com/thanos-io/thanos/blob/master/docs/components/sidecar.md),
or setting up third-party systems such as [Consul](https://github.com/cortexproject/cortex/issues/157), [Cassandra](https://cortexmetrics.io/docs/production/cassandra/),
[DynamoDB](https://cortexmetrics.io/docs/production/aws/) or [Memcached](https://cortexmetrics.io/docs/production/caching/).
- VictoriaMetrics has simpler architecture. This means less bugs and more useful features in the long run comparing to competing TSDBs.
See [comparing Thanos to VictoriaMetrics cluster](https://medium.com/@valyala/comparing-thanos-to-victoriametrics-cluster-b193bea1683)
and [Remote Write Storage Wars](https://promcon.io/2019-munich/talks/remote-write-storage-wars/) talk from [PromCon 2019](https://promcon.io/2019-munich/talks/remote-write-storage-wars/).
VictoriaMetrics also [uses less RAM than Thanos components](https://github.com/thanos-io/thanos/issues/448).
- Both systems support multi-tenancy out of the box. See [the corresponding docs for VictoriaMetrics](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#multitenancy).
- Both systems support data replication. See [replication in Cortex](https://github.com/cortexproject/cortex/blob/fe56f1420099aa1bf1ce09316c186e05bddee879/docs/architecture.md#hashing) and [replication in VictoriaMetrics](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#replication-and-data-safety).
- Both systems scale horizontally to multiple nodes. See [these docs](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#cluster-resizing-and-scalability) for details.
- Both systems support alerting and recording rules via the corresponding tools such as [vmalert](https://docs.victoriametrics.com/vmalert.html).
- Cortex may lose up to 12 hours of recent data on Ingestor failure - see [the corresponding docs](https://github.com/cortexproject/cortex/blob/fe56f1420099aa1bf1ce09316c186e05bddee879/docs/architecture.md#ingesters-failure-and-data-loss).
- Cortex is usually slower and requires more CPU and RAM than VictoriaMetrics. See [this talk from adidas at PromCon 2019](https://promcon.io/2019-munich/talks/remote-write-storage-wars/) and [other case studies](https://docs.victoriametrics.com/CaseStudies.html).
- VictoriaMetrics accepts data in multiple popular data ingestion protocols additionally to Prometheus remote_write protocol - InfluxDB, OpenTSDB, Graphite, CSV, JSON, native binary.
### What is the difference between VictoriaMetrics and [Thanos](https://github.com/thanos-io/thanos)?
- Thanos re-uses Prometheus source code, while VictoriaMetrics is written from scratch.
- VictoriaMetrics accepts data via [standard remote_write API for Prometheus](https://prometheus.io/docs/practices/remote_write/),
while Thanos uses non-standard [Sidecar](https://github.com/thanos-io/thanos/blob/master/docs/components/sidecar.md), which must run alongside each Prometheus instance.
- Thanos Sidecar requires disabling data compaction in Prometheus, which may hurt Prometheus performance and increase RAM usage. See [these docs](https://thanos.io/components/sidecar.md/) for more details.
- Thanos stores data in object storage (Amazon S3 or Google GCS), while VictoriaMetrics stores data in block storage
([GCP persistent disks](https://cloud.google.com/compute/docs/disks#pdspecs), Amazon EBS or bare metal HDD).
While object storage is usually less expensive, block storage provides much lower latencies and higher throughput.
VictoriaMetrics works perfectly with HDD-based block storage - there is no need in using more expensive SSD or NVMe disks in most cases.
- Thanos may lose up to 2 hours of recent data, which wasn't uploaded yet to object storage. VictoriaMetrics may lose only a few seconds of recent data,
which isn't synced to persistent storage yet. See [this article for details](https://medium.com/@valyala/wal-usage-looks-broken-in-modern-time-series-databases-b62a627ab704).
- Thanos may be harder to setup and operate comparing to VictoriaMetrics, since it has more moving parts, which can be connected with less reliable networks.
See [this article for details](https://medium.com/faun/comparing-thanos-to-victoriametrics-cluster-b193bea1683).
- Thanos is usually slower and requires more CPU and RAM than VictoriaMetrics. See [this talk from adidas at PromCon 2019](https://promcon.io/2019-munich/talks/remote-write-storage-wars/).
- VictoriaMetrics accepts data in multiple popular data ingestion protocols additionally to Prometheus remote_write protocol - InfluxDB, OpenTSDB, Graphite, CSV, JSON, native binary.
- VictoriaMetrics requires [10x less RAM](https://medium.com/@valyala/insert-benchmarks-with-inch-influxdb-vs-victoriametrics-e31a41ae2893) and it [works faster](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae).
- VictoriaMetrics provides [better query language](https://medium.com/@valyala/promql-tutorial-for-beginners-9ab455142085) than InfluxQL or Flux.
- VictoriaMetrics accepts data in multiple popular data ingestion protocols additionally to InfluxDB - Prometheus remote_write, OpenTSDB, Graphite, CSV, JSON, native binary.
- TimescaleDB insists on using SQL as a query language. While SQL is more powerful than PromQL, this power is rarely required during typical TSDB usage. Real-world queries usually [look clearer and simpler when written in PromQL than in SQL](https://medium.com/@valyala/promql-tutorial-for-beginners-9ab455142085).
- VictoriaMetrics requires [up to 70x less storage space comparing to TimescaleDB](https://medium.com/@valyala/when-size-matters-benchmarking-victoriametrics-vs-timescale-and-influxdb-6035811952d4) for storing the same amount of time series data. The gap in storage space usage can be lowered from 70x to 3x if [compression in TimescaleDB is properly configured](https://docs.timescale.com/latest/using-timescaledb/compression) (it isn't an easy task in general case :)).
- VictoriaMetrics accepts data in multiple popular data ingestion protocols - InfluxDB, OpenTSDB, Graphite, CSV, while TimescaleDB supports only SQL inserts.
### Does VictoriaMetrics use Prometheus technologies like other clustered TSDBs built on top of Prometheus such as [Thanos](https://github.com/thanos-io/thanos) or [Cortex](https://github.com/cortexproject/cortex)?
No. VictoriaMetrics core is written in Go from scratch by [fasthttp](https://github.com/valyala/fasthttp) [author](https://github.com/valyala).
The architecture is [optimized for storing and querying large amounts of time series data with high cardinality](https://medium.com/devopslinks/victoriametrics-creating-the-best-remote-storage-for-prometheus-5d92d66787ac). VictoriaMetrics storage uses [certain ideas from ClickHouse](https://medium.com/@valyala/how-victoriametrics-makes-instant-snapshots-for-multi-terabyte-time-series-data-e1f3fb0e0282). Special thanks to [Alexey Milovidov](https://github.com/alexey-milovidov).
* [Prometheus vs VictoriaMetrics benchmark on node-exporter metrics](https://valyala.medium.com/prometheus-vs-victoriametrics-benchmark-on-node-exporter-metrics-4ca29c75590f)
* [Promscale vs VictoriaMetrics: measuring resource usage in production](https://valyala.medium.com/promscale-vs-victoriametrics-resource-usage-on-production-workload-91c8e3786c03)
* [Benchmarking time series workloads on Apache Kudu using TSBS](https://blog.cloudera.com/benchmarking-time-series-workloads-on-apache-kudu-using-tsbs/)
* [Billy: how VictoriaMetrics deals with more than 500 billion rows](https://medium.com/@valyala/billy-how-victoriametrics-deals-with-more-than-500-billion-rows-e82ff8f725da)
* [Measuring vertical scalability for time series databases: VictoriaMetrics vs InfluxDB vs TimescaleDB](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae).
* [Measuring insert performance on high-cardinality time series: VictoriaMetrics vs InfluxDB](https://medium.com/@valyala/insert-benchmarks-with-inch-influxdb-vs-victoriametrics-e31a41ae2893)
* [TSBS benchmark on high-cardinality time series: VictoriaMetrics vs InfluxDB vs TimescaleDB](https://medium.com/@valyala/high-cardinality-tsdb-benchmarks-victoriametrics-vs-timescaledb-vs-influxdb-13e6ee64dd6b)
* [Standard TSBS benchmark: VictoriaMetrics vs InfluxDB vs TimescaleDB](https://medium.com/@valyala/when-size-matters-benchmarking-victoriametrics-vs-timescale-and-influxdb-6035811952d4)
### Why VictoriaMetrics doesn't support [Prometheus remote read API](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#%3Cremote_read%3E)?
Remote read API requires transferring all the raw data for all the requested metrics over the given time range. For instance,
if a query covers 1000 metrics with 10K values each, then the remote read API had to return `1000*10K`=10M metric values to Prometheus.
This is slow and expensive.
Prometheus remote read API isn't intended for querying foreign data aka `global query view`. See [this issue](https://github.com/prometheus/prometheus/issues/4456) for details.
### Does VictoriaMetrics fit for data from IoT sensors and industrial sensors?
VictoriaMetrics is able to handle data from hundreds of millions of IoT sensors and industrial sensors.
It supports [high cardinality data](https://medium.com/@valyala/high-cardinality-tsdb-benchmarks-victoriametrics-vs-timescaledb-vs-influxdb-13e6ee64dd6b),
perfectly [scales up on a single node](https://medium.com/@valyala/measuring-vertical-scalability-for-time-series-databases-in-google-cloud-92550d78d8ae)
and scales horizontally to multiple nodes.
### Where can I ask questions about VictoriaMetrics?
