Run MongoDB on Arm-based Azure Cobalt 200 instances
Introduction
What are Cobalt 200 and MongoDB?
Create an Arm-based cloud virtual machine using Cobalt 200
Install MongoDB and Mongosh
MongoDB baseline testing
Generate a MongoDB workload
Monitor MongoDB with mongotop
Monitor MongoDB with mongostat
Next Steps
Run MongoDB on Arm-based Azure Cobalt 200 instances
Observe server activity
Keep the workload running in terminal 1. In terminal 3, SSH into the same Cobalt 200 VM and collect three samples at two-second intervals:
mongostat --host 127.0.0.1 --port 27017 --rowcount=3 2
The command stops after three samples. To monitor continuously, omit --rowcount=3 and press Ctrl+C to stop.
The output is similar to:
insert query update delete getmore command dirty used flushes vsize res qrw arw net_in net_out conn time
1894 1894 1894 1894 0 1|0 0.5% 0.6% 0 3.55G 157M 0|0 0|0 1.31m 607k 10 Nov 10 18:58:39.089
1867 1867 1867 1867 0 0|0 0.6% 0.7% 0 3.55G 157M 0|0 0|0 1.29m 599k 10 Nov 10 18:58:41.089
1831 1831 1831 1831 0 1|0 0.7% 0.9% 0 3.55G 159M 0|0 0|0 1.27m 589k 11 Nov 10 18:58:43.090
These samples were collected on a Standard_D4ps_v7 Cobalt 200 VM (4 vCPUs, 16 GiB) running Ubuntu 24.04 LTS, MongoDB 8.0.12, and Database Tools 100.13.0 with the provided single-client workload. Your results will vary.
Interpret the metrics
- insert, query, update, delete report operation rates. Some modern read operations are reported as commands rather than legacy query operations.
- getmore reports operations that fetch additional cursor results.
- command reports database command activity, including commands issued by the monitoring tools.
- dirty/used describe dirty and used portions of the WiredTiger cache.
- vsize/res report virtual and resident process memory.
- qrw/arw describe queued and active readers and writers.
- net_in/net_out show incoming and outgoing network traffic rates.
- conn reports active connections.
- time identifies each sample.
For field definitions, see the mongostat documentation .
Evaluate the baseline
The workload performs inserts, reads, updates, and deletes, so you should see activity while it runs. Compare several samples and use mongotop to identify the active collection. If activity is absent, check whether the 60-second workload has already finished and re-run it if needed.
These short samples demonstrate monitoring of a functional local MongoDB deployment on Cobalt 200. They do not establish sustained throughput, request latency, production capacity, or a performance or cost comparison with other VM types. Use representative data and a separate load generator for capacity testing.