A presentation at SnowCamp 2019 in in Grenoble, France by Horacio Gonzalez
Monitoring OVH 300k servers, 27 DCs… and one Metrics platform Horacio Gonzalez @LostInBrittany Monitoring @LostInBrittany
Sommaire temporaire - Intro we and OVH (5 minutes) - Intro our talk (2 minutes) - Make Better Decisions By using Numbers (5 minutes) - Building OVH Metrics (10 minutes) - Conclusion (2 minutes) - Bye bye (1 minute) Monitoring @LostInBrittany
Who are we? Introducing myself and introducing OVH Monitoring @LostInBrittany
Horacio Gonzalez @LostInBrittany Spaniard lost in Brittany, developer, dreamer and all-around geek Monitoring @LostInBrittany
OVH : Key Figures 1.3M Customers worldwide in 138 Countries 1.5 Billions euros investment over five years 30 Datacenters (growing) 350k Dedicated Servers 200k Private cloud VMs running 650k Public cloud Instances created in a month 15TB bandwidth capacity
OVH: A Global Leader on Cloud 200k Private cloud VMs running 1 Dedicated IaaS Europe 2018 27 Datacenters Own 15 Tbps Hosting capacity : 1.3M Physical Servers 360k Servers already deployed Netwok with 35 PoPs 2020 50 Datacenters
1.3M Customers in 138 Countries Monitoring @LostInBrittany
Ranking & Recognition 1st European Cloud Provider* 1st Hosting provider in Europe 1st Provider Microsoft Exchange Certified vCloud Datacenter Certified Kubernetes platform (CNCF) Vmware Global Service Provider 2013-2016 Veeam Best Cloud Partner of the year (2018) Monitoring @LostInBrittany
OVH: Our solutions Cloud Web Hosting Mobile Hosting Telecom VPS Containers ▪ Dedicated Server Domain names VoIP Public Cloud Compute ▪ Data Storage Email SMS/Fax Private Cloud ▪ Network and Database CDN Virtual desktop Serveur dédié Security Object Storage Web hosting Cloud HubiC Over theBox ▪ Licences Cloud Desktop Securities MS Office Hybrid Cloud Messaging MS solutions Monitoring @LostInBrittany
Once upon a time… Because I love telling tales Monitoring @LostInBrittany
This talk is about a tale… A true one nevertheless Monitoring @LostInBrittany
And as in most tales It begins with a mission Monitoring @LostInBrittany
And a band of heroes Engulfed into the adventure Monitoring @LostInBrittany
They fight against mishaps And all kind of foes Monitoring @LostInBrittany
They build mighty fortresses Pushing the limits of possible Monitoring @LostInBrittany
And defend them day after day Against all odds Monitoring @LostInBrittany
But we don’t know yet the end Because this tale isn’t finished yet Monitoring @LostInBrittany
It begins with a mission Build a metrics platform for OVH Monitoring @LostInBrittany
Why do we need metrics? To make better decisions by using numbers Monitoring @LostInBrittany
Why do we need metrics? We want our code to add value Monitoring @LostInBrittany
Why do we need metrics? We need to make better decisions about our code Monitoring @LostInBrittany
Why do we need metrics? Code adds value when it runs not when we write it Monitoring @LostInBrittany
Why do we need metrics? We need to know what our code does when it runs Monitoring @LostInBrittany
Why do we need metrics? We can’t do this unless we measure it Monitoring @LostInBrittany
Why do we need metrics? We have a mental model of what our code does Monitoring @LostInBrittany
Why do we need metrics? This representation can be wrong Monitoring @LostInBrittany
Why do we need metrics? We can’t know until we measure it Monitoring @LostInBrittany
Find the bottleneck ‘’ “The app is slow.” - User Monitoring @LostInBrittany
Find the bottleneck ‘’ “The app is slow.” - User “The page takes 500ms!” - Ops Monitoring @LostInBrittany
Find the bottleneck ? SQL Query? Template Rendering? Session Storage? Monitoring @LostInBrittany
Find the bottleneck ? We don’t know Monitoring @LostInBrittany
With observability: SQL Query………………………….53ms Template Rendering……….1ms Session Storage……………315ms Monitoring @LostInBrittany
With observability: SQL Query………………………….53ms Template Rendering……….1ms Session Storage……………315ms Monitoring @LostInBrittany
Why do we need metrics? We improve our mental model by measuring what our code does Monitoring @LostInBrittany
Why do we need metrics? We use our mental model to decide what to do Monitoring @LostInBrittany
Why do we need metrics? A better mental model makes us better at deciding what to do Monitoring @LostInBrittany
Why do we need metrics? Better decisions makes us better at generating value Monitoring @LostInBrittany
Why do we need metrics? Measuring make your App better Monitoring @LostInBrittany
It began with a mission Build a metrics platform for OVH Monitoring @LostInBrittany
A metrics platform for OVH For all OVH Monitoring @LostInBrittany
Building OVH Metrics One Platform to unify them all, One Platform to find them, One Platform to bring them all and in the Metrics monitor them Monitoring @LostInBrittany
What is OVH Metrics? Managed Cloud Platform for Time Series Monitoring @LostInBrittany
OVH monitoring story We had lots of partial solutions… Monitoring @LostInBrittany
OVH monitoring story One Platform to unify them all What should we build it on? Monitoring @LostInBrittany
OVH monitoring story Including a really big Monitoring @LostInBrittany
OpenTSDB drawbacks OpenTSDB RowKey Design ! Monitoring @LostInBrittany
OpenTSDB Rowkey design flaws ● .regex. => full table scans ● High cardinality issues (Query latencies) We needed something able to manage hundreds of millions time series OpenTSBD didn’t scale for us Monitoring @LostInBrittany
OpenTSDB other flaws ● ● ● ● ● Compaction (or append writes) /api/query : 1 endpoint per function? Asynchronous Unauthenticated … Monitoring @LostInBrittany
Scaling OpenTSDB Monitoring @LostInBrittany
Metrics needs First need: To be massively scalable Monitoring @LostInBrittany
Analytics is the key to success Fetching data is only the tip of the iceberg Monitoring @LostInBrittany
Analysing metrics data To be scalable, analysis must be done in the database, not in user’s computer Monitoring @LostInBrittany
Metrics needs Second need: To have rich query capabilities Monitoring @LostInBrittany
Enter Warp 10… Open-source Time series Database Monitoring @LostInBrittany
More than a Time Series DB Warp 10 is a software platform that ● Ingests and stores time series ● Manipulates and analyzes time series Monitoring @LostInBrittany
Manipulating Time Series with Warp 10 A true Time Series analysis toolbox ○ Hundreds of functions ○ Manipulation frameworks ○ Analysis workflow Monitoring @LostInBrittany
Manipulating Time Series with Warp 10 A Time Series manipulation language WarpScript Monitoring @LostInBrittany
Did you say scalability? From the smallest to the largest… Monitoring @LostInBrittany
More Warp 10 goodness ● Secured & multi tenant ● Synchronous (transactions) ● In memory Index ● Better Performance ● No cardinality issues ● Better Scalability ● Lockfree ingestion ● Versatile ● WarpScript Query Language (standalone, distributed) ● Support more data types Monitoring @LostInBrittany
Metrics Data Platform + + Monitoring @LostInBrittany
Metrics Data Platform Monitoring @LostInBrittany
Building an ecosystem From Warp 10 to OVH Metrics Monitoring @LostInBrittany
Multi-protocol Why to choose? We need them all! Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Monitoring @LostInBrittany
Open source monitoring tools Why choose? Let’s support all of them! Monitoring @LostInBrittany
Metrics Platform Monitoring @LostInBrittany
Metrics Platform graphite influx https:// opentsdb .<region>.metrics.ovh.net prometheus warp10 … Monitoring @LostInBrittany
Metrics Live In-memory, high-performance Metrics instances Monitoring @LostInBrittany
In-memory: Metrics live +120 million of writes/s Monitoring @LostInBrittany
In-memory: Metrics live Monitoring @LostInBrittany
In-memory: Metrics live Monitoring @LostInBrittany
Monitoring is only the beginning OVH Metrics answer to many other use cases Monitoring @LostInBrittany
Use cases families • • • • Billing Monitoring IoT (e.g. bill on monthly max consumption) ……………………………………………..……. (APM, infrastructure,appliances,…) …..…………………………… (Manage devices, operator integration, …) …………………………………………….…………………. Geo Location (Manage localized fleets) ……..………………… Monitoring @LostInBrittany
Use cases • • • • • • DC Temperature/Elec/Cooling map Pay as you go billing (PCI/IPLB) GSCAN Monitoring ML Model scoring (Anti-Fraude) Pattern Detection for medical applications Monitoring @LostInBrittany
SREing Metrics With a great power comes a great responsibility Monitoring @LostInBrittany
Metrics’ own metrics 432 000 000 000 datapoints / day Monitoring @LostInBrittany
Metrics’ own metrics 10 Tb / day Monitoring @LostInBrittany
Metrics’ own metrics 5 000 000 dp/s Monitoring @LostInBrittany
Metrics’ own metrics 500 000 000 series Monitoring @LostInBrittany
Our clusters size GRA: BHS: ● 150 nodes ● 2 PB ● 1.1 Gbps ● 30 nodes ● 400 TB ● 120 Mbps Monitoring @LostInBrittany
Our cluster architecture Warp10 Ingress Warp10 Warp10 Directory Directory Kafka Warp10 Warp10 Egress Egress Warp10 Warp10 Store Store Region server + Datanode Region server + Datanode Region server + Datanode Monitoring Region server + Datanode @LostInBrittany
Detecting errors Before it’s too late Monitoring 86 @LostInBrittany
Extract errors from logs Monitoring @LostInBrittany
Tailor Forward logs and extract metrics! Monitoring @LostInBrittany
Monitoring the JVM Monitoring @LostInBrittany
Documentation Monitoring @LostInBrittany
JVM GC The good, the bad and the ugly Monitoring @LostInBrittany
The good Monitoring @LostInBrittany
The bad Monitoring @LostInBrittany
… and the ugly #java #jdk11 #zgc Monitoring @LostInBrittany
Monitoring HBase Monitoring @LostInBrittany
Number of open regions Monitoring @LostInBrittany
Queues length Monitoring @LostInBrittany
Number of read and write requests Monitoring @LostInBrittany
Preserve data locality Monitoring @LostInBrittany
Host health Monitoring @LostInBrittany
Pokédex Inventory all animals. Monitoring @LostInBrittany
Merging all data sources Monitoring @LostInBrittany
Global visualization Monitoring @LostInBrittany
Correlate information Monitoring @LostInBrittany
Sacha The best tamer Monitoring @LostInBrittany
An awesome CLI Monitoring @LostInBrittany
Retrieving bare informations Monitoring @LostInBrittany
Create region map Monitoring @LostInBrittany
Move region to another region server Monitoring @LostInBrittany
Drain regions of the region server Monitoring @LostInBrittany
Managing multiple hardware profiles Monitoring @LostInBrittany
Balance the cluster Monitoring @LostInBrittany
Conclusion That’s all folks! Monitoring @LostInBrittany
What to do when you must monitor the whole infrastructure of the biggest European hosting and cloud provider? How to choose a tool when the most used ones fail to scale to your needs? How to build an Metrics platform to unify, conciliate and replace years of fragmented legacy partial solutions?
In this talk we will relate our experience building and maintaining OVH Metrics, the platform used to monitor all OVH infrastructure. We needed to go to places where most monitoring solutions hadn’t gone before, it needed to operate at the scale of the biggest European hosting and cloud providers: 27 data centers, more than 300k servers (bare metal!), and hundreds of products to fulfill our mission to host 1.3 million customers.
You will hear about time series, about open source solutions pushed to the limit, about HBase clusters operated at the extreme, and how about a small team leveraged the power of a handful of open source solution and lots of coding glue to build one of the most performant monitoring solutions ever.
The following resources were mentioned during the presentation or are useful additional information.