Sysadmins, site reliability engineers (SREs), and cloud operators all too typically battle to really feel assured of their infrastructure because it scales up. Also too typically, they suppose the one technique to remedy their challenges is to jot down a instrument for in-house use. Fortunately, there are alternatives. There are many open supply instruments out there to check an infrastructure’s efficiency. Here are my favorites.
Pbench is a efficiency testing harness to make executing benchmarks and efficiency instruments simpler and extra handy. In brief, it:
- Excels at operating micro-benchmarks on giant scales of hosts (bare-metal, digital machines, containers, and so on.) whereas automating a probably giant set of benchmark parameters
- Focuses on putting in, configuring, and executing benchmark code and efficiency instruments and never on provisioning or orchestrating the testbed (e.g., OpenStack, RHEV, RHEL, Docker, and so on.)
- Is designed to work in live performance with provisioning instruments like BrowBeat or Ansible playbooks
Baselining is a vital facet of infrastructure reliability. Ripsaw is a efficiency benchmark Operator for launching workloads on Kubernetes. It deploys as a Kuberentes Operator that then deploys widespread workloads, together with particular purposes (e.g., Couchbase) or basic efficiency exams (e.g., Uperf) to measure and set up a efficiency baseline.
The assortment of instruments in OpenShift Scale, OpenShift’s open supply answer for efficiency testing, do the whole lot from spinning up OpenShift on OpenStack installations (TripleO Install and ShiftStack Install), putting in on Amazon Web Services (AWS), or offering containerized tooling, like operating Pbench in your cluster or doing cluster limits testing, community exams, storage exams, metric exams with Prometheus, logging, and concurrent construct testing.
Scale’s CI suite is versatile sufficient to each add workloads and embrace your workloads when deploying to Azure or anyplace else you may run. You can see the complete suite of instruments on GitHub.
Browbeat calls itself “a performance tuning and analysis tool for OpenStack.” You can use it to research and tune the deployment of your workloads. It additionally automates the deployment of ordinary monitoring and information evaluation instruments like Grafana and Graphite. Browbeat is maintained on GitHub.
Smallfile is a filesystem workload generator focused for scale-out, distributed storage. It has been used to check plenty of open filesystem applied sciences, together with GlusterFS, CephFS, Network File System (NFS), Server Message Block (SMB), and OpenStack Cinder volumes. It is maintained on GitHub.
Ceph Benchmarking Tool (CBT) is a testing harness that may automate duties for testing Ceph cluster efficiency. It information system metrics with collectl, and it might probably gather extra data with instruments together with perf, blktrace, and valgrind. CBT can even do superior testing that features automated object storage daemon outages, erasure-coded swimming pools, and cache-tier configurations.
Contributors have prolonged CBT to make use of Pbench monitoring tools and Ansible and to run the Smallfile benchmark. A separate Grafana visualization dashboard makes use of Elasticsearch information generated by Automated Ceph Test.
Satellite-performance (satperf) is a set of Ansible playbooks and helper scripts to deploy Satellite 6 environments and measure the efficiency of chosen actions, equivalent to concurrent registrations, distant execution, Puppet operations, repository synchronizations and promotions, and extra. You can discover Satperf on GitHub.
Sysadmins, SREs, and cloud operators face all kinds of challenges as they work to scale their infrastructure, however fortunately there’s additionally all kinds of instruments to assist them get previous these widespread points. Any of those seven instruments ought to enable you get began testing your infrastructure’s efficiency because it scales.
Are there different open supply efficiency and scaling instruments that must be on this record? Add your favorites within the feedback.