TL
The short answer

Azure B series virtual machines are burstable instances that bill for a low guaranteed baseline CPU level and accumulate CPU credits whenever usage sits below that baseline, which can then be spent to burst toward full core performance for limited periods. They are the cheapest fit for workloads that are idle or light most of the time and spike occasionally, such as small web servers, development boxes, low traffic services, and test environments. They are the wrong fit, and quietly more expensive in real terms, for workloads that run consistently busy, because once credits are exhausted performance is throttled to the baseline and you either suffer slow response or pay to size up anyway. The decision is the CPU utilization profile, not the sticker price.

Here is how the credit mechanism works and how to decide whether burstable saves you money.

How do B series CPU credits work?

Each B series size has a baseline CPU percentage it is guaranteed and billed around. When the virtual machine runs below that baseline it banks credits up to a cap; when it runs above the baseline it spends banked credits to reach higher performance. While credits last, an idle then bursty workload gets near full core performance for a fraction of the cost of a same sized steady instance. When credits run out, the machine is throttled back to its baseline until it banks more. So the model rewards a duty cycle that is mostly quiet with short bursts, and penalises sustained load that drains credits and then runs throttled.

When does burstable actually save money?

Workload profileBurstable fitWhy
Idle most of the time, occasional spikesStrongBanks credits while idle, spends them on bursts
Low traffic web or microserviceStrongBaseline covers normal load cheaply
Development and test boxesStrongOften idle, rarely sustained
Steady moderate to high CPUPoorDrains credits, then throttles to baseline
Batch or compute heavy jobsPoorSustained load needs a steady or compute size

The test is simple: if average CPU sits comfortably below the size baseline with short spikes, burstable wins; if it runs consistently near or above baseline, a steady general purpose or compute optimised size is cheaper per unit of real work. Pricing and baseline percentages are indicative; verify against current Azure pricing.

How does burstable interact with rightsizing and reservations?

Burstable is a rightsizing tool, so it belongs in the rightsizing step before any commitment. Move genuinely light workloads onto an appropriately sized B series, move consistently busy ones onto a steady size, and only then reserve the steady floor. Reservations and the Azure savings plan apply to B series too, so a stable burstable fleet can still be committed, but the bigger win is matching the duty cycle to the family first. Watching CPU credit balance over time tells you whether a machine is correctly placed: chronically zero credits means the workload outgrew burstable and should move to a steady size.

A worked example

Worked example

A European SaaS company ran dozens of low traffic internal services on steady general purpose virtual machines sized for rare peaks, so they idled most of the day while billing full rate. Profiling CPU showed average utilization far below baseline with short spikes, the ideal burstable pattern. Moving those services to appropriately sized B series let them bank credits while idle and burst during spikes at a fraction of the steady cost, while a handful of consistently busy services that had been wrongly placed on burstable were moved the other way onto steady sizes. Matching family to duty cycle cut the compute line with no performance regression, verified against billing data and anonymised.

Frequently asked questions

What are Azure B series VMs good for?
Workloads that are idle or light most of the time and burst occasionally, such as small web servers, low traffic services, and development or test boxes. They bank CPU credits while below baseline and spend them during short bursts, which makes them cheaper than a steady size for that profile.
When are burstable VMs a false economy?
When a workload runs consistently busy. Once banked CPU credits are exhausted the machine is throttled to its baseline, so you either accept slow performance or size up anyway. Steady moderate to high CPU and batch jobs belong on general purpose or compute optimised sizes.
How do I know if a VM should be burstable?
Profile its CPU. If average utilization sits comfortably below the size baseline with short spikes, burstable wins. If it runs near or above baseline, or its credit balance is chronically zero, move it to a steady size that is cheaper per unit of real work.

Match every VM family to its duty cycle

We help enterprises profile workloads and place each on the cheapest virtual machine family for its real duty cycle, then commit the steady floor, as an independent advisory that takes zero provider commissions and answers only to you. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on a Fixed Fee or a no risk Gainshare basis. Download the Azure commitment kit, read the deeper Azure cost optimization guide, and see VM rightsizing that engineers accept.

Independent · buyer-side

Put a defensible number on your cloud spend.

No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.

Buyer-side intelligence, monthly.

The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.