Linux Virtual Machines Pricing
- No upfront cost
- No termination fees
- Per-second billing
Start your Azure free account and get 12 months free access to Virtual Machines plus $200 credit for 30 days.
Azure Virtual Machines gives you the flexibility of virtualisation for a wide range of computing solutions with support for Linux, Windows Server, SQL Server, Oracle, IBM, SAP and more. All current generation Virtual Machines include load balancing and auto-scaling at no cost. For optimal performance, we recommend pairing your Virtual Machines with Managed Disks. Standard egress charges apply.
Choose from the following payment options for Virtual Machines. Combine either option with the Azure Hybrid Benefit for additional savings.
Pay as you go
Pay for compute capacity by the second, with no long-term commitment or upfront payments. Increase or decrease compute capacity on demand. Start or stop at any time and only pay for what you use.
Recommended for:
- Users who prefer the low cost and flexibility of Azure Virtual Machines.
- Applications with short-term, spiky or unpredictable workloads that cannot be interrupted.
- Applications being developed or tested on Azure Virtual Machines for the first time.
Reserved Virtual Machine Instances
An Azure Reserved Virtual Machine Instance is an advanced purchase of a Virtual Machine for one or three years in a specified region. The commitment is made up front and in return, you get up to 72 percent price savings compared to pay-as-you-go pricing. Reserved Virtual Machine Instances are flexible and can easily be exchanged or returned.
Recommended for:
- Applications with steady-state usage.
- Customers who want budget predictability.
- Customers who can commit to using a virtual machine (VM) over a one-year or three-year term to reduce computing costs.
Example Virtual Machines pricing
Azure Virtual Machines are designed for all budgets and workloads, starting from our economical B-series to our newest GPU VMs optimised for machine learning. Explore popular categories:
Burstable VMs—B1S
Most economical
Our B-series VMs provide an economical, low-cost solution for workloads that normally don't use a lot of CPU, but occasionally need to burst to handle higher workloads. Free for 12 months.
Compute optimised—Fsv2
Raw compute power
Fv2 is our newest compute-optimised VM family and uses the Intel Skylake processor. Fv2 delivers the latest Intel CPU for raw compute power.
General purpose—Dv3
Balanced CPU and memory
Our Dv3 family is the latest generation of our general purpose VMs. It is appropriate for a variety of workloads.
Memory optimised—Ev3
High memory-to-core ratio
Ev3 is our latest generation memory-optimised VM. It is great for relational database servers, caches, and in-memory analytics.
ND VMs now available
Designed for artificial intelligence (AI) and deep learning
- Powered by NVIDIA Tesla P40 GPUs.
- Ideal for AI workloads using CNTK, TensorFlow, Caffe and other frameworks.
- Larger GPU memory sizes of up to 24 GB available.
- RDMA and InfiniBand connectivity for large-scale training jobs.
Explore all Virtual Machine options
Explore options in Azure Virtual Machines. Choose your OS and software combination and find VMs for your vCPU (core), RAM and storage requirements.
US government entities are eligible to purchase Azure Government services from a licensing solution provider with no upfront financial commitment or directly through a pay-as-you-go online subscription.
Important—The price in R$ is merely a reference; this is an international transaction and the final price is subject to exchange rates and the inclusion of IOF taxes. An eNF will not be issued.
Azure Germany is available to customers and partners doing business in the European Union (EU), the European Free Trade Association (EFTA) and in the United Kingdom (UK). It provides data residency in Germany with additional levels of control and data protection. You can also sign up for a free Azure Germany trial.
General purpose
Balanced CPU-to-memory ratio. Ideal for testing and development, small to medium databases and low to medium traffic web servers.
B-series
B-series are economical virtual machines that provide a low-cost option for workloads that typically run at a low to moderate baseline CPU performance, but sometimes need to burst to significantly higher CPU performance when the demand rises. These workloads do not require the use of the full CPU all the time, but occasionally will need to burst to finish some tasks more quickly. Many applications such as development and test servers, low traffic web servers, small databases, micro services, servers for proof-of-concepts, build servers and code repositories fit into this model.| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| B1S | 1 | 1.00 GiB | 2 GiB | $- | $- | $- | |
| B2S | 2 | 4.00 GiB | 8 GiB | $- | $- | $- | |
| B1MS | 1 | 2.00 GiB | 4 GiB | $- | $- | $- | |
| B2MS | 2 | 8.00 GiB | 16 GiB | $- | $- | $- | |
| B4MS | 4 | 16.00 GiB | 32 GiB | $- | $- | $- | |
| B8MS | 8 | 32.00 GiB | 64 GiB | $- | $- | $- |
Av2 Standard
Av2 Standard is the latest generation of A-series virtual machines with similar CPU performance and faster disk. These virtual machines are suitable for development workloads, build servers, code repositories, low-traffic websites and web applications, micro services, early product experiments and small databases. Like the prior A Standard generation, Av2 virtual machines will include load balancing and auto-scaling at no additional charge.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| A1 v2 | 1 | 2.00 GiB | 10 GiB | $- | – – | – – | |
| A2 v2 | 2 | 4.00 GiB | 20 GiB | $- | – – | – – | |
| A4 v2 | 4 | 8.00 GiB | 40 GiB | $- | – – | – – | |
| A8 v2 | 8 | 16.00 GiB | 80 GiB | $- | – – | – – | |
| A2m v2 | 2 | 16.00 GiB | 20 GiB | $- | – – | – – | |
| A4m v2 | 4 | 32.00 GiB | 40 GiB | $- | – – | – – | |
| A8m v2 | 8 | 64.00 GiB | 80 GiB | $- | – – | – – |
D2-64 v3 latest generation
D2-64 v3 instances are the latest hyper-threaded generation of general purpose instances. D2-64 v3 instances are based on the 2.3 GHz Intel XEON ® E5-2673 v4 (Broadwell) processor and can achieve 3.5 gigahertz (GHz) with Intel Turbo Boost Technology 2.0. D2-64 v3 instances offer the combination of CPU, memory, and local disk for most production workloads.
| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| D2 v3 | 2 | 8.00 GiB | 50 GiB | $- | $- | $- | |
| D4 v3 | 4 | 16.00 GiB | 100 GiB | $- | $- | $- | |
| D8 v3 | 8 | 32.00 GiB | 200 GiB | $- | $- | $- | |
| D16 v3 | 16 | 64.00 GiB | 400 GiB | $- | $- | $- | |
| D32 v3 | 32 | 128.00 GiB | 800 GiB | $- | $- | $- | |
| D64 v3 | 64 | 256.00 GiB | 1,600 GiB | $- | $- | $- |
D2s-64s v3 latest generation
Dsv3-series support premium storage and are the latest hyper-threaded generation of general purpose instances running on both the Intel Xenon E5-2673 v3 2.4GHz (Haswell) and the Intel Xenon ® E5-2673 v4 2.3GHz (Broadwell) processor. With the Intel Turbo Boost Technology 2.0 the Dsv3 can go up to 3.1GHz. The Dsv3-series sizes offer a combination of vCPU, memory, and temporary storage for most production workloads.
The VM pricing and billing meters for Dsv3 sizes are the same as Dv3-series. Persistent storage disks are billed separately from virtual machines. See pricing for disks.
| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| D2s v3 | 2 | 8.00 GiB | 16 GiB | $- | $- | $- | |
| D4s v3 | 4 | 16.00 GiB | 32 GiB | $- | $- | $- | |
| D8s v3 | 8 | 32.00 GiB | 64 GiB | $- | $- | $- | |
| D16s v3 | 16 | 64.00 GiB | 128 GiB | $- | $- | $- | |
| D32s v3 | 32 | 128.00 GiB | 256 GiB | $- | $- | $- | |
| D64s v3 | 64 | 256.00 GiB | 512 GiB | $- | $- | $- |
D1-5 v2
D1-5 v2 instances are based on the 2.4 GHz Intel Xeon® E5-2673 v3 (Haswell) processor and can achieve 3.1 GHz with Intel Turbo Boost Technology 2.0. D1-5 v2 instances offer a powerful combination of CPU, memory and local disk for most production applications.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| D1 v2 | 1 | 3.50 GiB | 50 GiB | $- | $- | $- | |
| D2 v2 | 2 | 7.00 GiB | 100 GiB | $- | $- | $- | |
| D3 v2 | 4 | 14.00 GiB | 200 GiB | $- | $- | $- | |
| D4 v2 | 8 | 28.00 GiB | 400 GiB | $- | $- | $- | |
| D5 v2 | 16 | 56.00 GiB | 800 GiB | $- | $- | $- |
DS1-S5 v2
Persistent storage disks are billed separately from virtual machines. To use premium storage disks, use the variant Dsv2 virtual machines. The pricing and billing meters for Dsv2 sizes are the same as Dv2-series. See pricing for disks.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| DS1 v2 | 1 | 3.50 GiB | 7 GiB | $- | $- | $- | |
| DS2 v2 | 2 | 7.00 GiB | 14 GiB | $- | $- | $- | |
| DS3 v2 | 4 | 14.00 GiB | 28 GiB | $- | $- | $- | |
| DS4 v2 | 8 | 28.00 GiB | 56 GiB | $- | $- | $- | |
| DS5 v2 | 16 | 56.00 GiB | 112 GiB | $- | $- | $- |
Compute optimised
High CPU-to-memory ratio. Good for medium traffic web servers, network appliances, batch processes and application servers.
Fsv2-series
The Fsv2-series virtual machines support 2-GiB RAM and 8 GB of local temporary storage (SSD) per vCPU, and are optimized for compute intensive workloads. The F-series is hyper-threaded and based on the 2.7 GHz Intel Xeon® Platinum 8168 (SkyLake) processor, which can achieve clock speeds as high as 3.7 GHz with the Intel Turbo Boost Technology 2.0. These virtual machines are suitable for scenarios like batch processing, web servers, analytics, and gaming.
| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| F2s v2 | 2 | 4.00 GiB | 16 GiB | $- | $- | $- | |
| F4s v2 | 4 | 8.00 GiB | 32 GiB | $- | $- | $- | |
| F8s v2 | 8 | 16.00 GiB | 64 GiB | $- | $- | $- | |
| F16s v2 | 16 | 32.00 GiB | 128 GiB | $- | $- | $- | |
| F32s v2 | 32 | 64.00 GiB | 256 GiB | $- | $- | $- | |
| F64s v2 | 64 | 128.00 GiB | 512 GiB | $- | $- | $- | |
| F72s v2 | 72 | 144.00 GiB | 576 GiB | $- | $- | $- |
F-series
The F-series virtual machines support 2-GiB RAM and 16 GB of local solid state drive (SSD) per CPU core and are optimised for compute intensive workloads. The F-series is based on the 2.4 GHz Intel Xeon® E5-2673 v3 (Haswell) processor, which can achieve clock speeds as high as 3.2 GHz with the Intel Turbo Boost Technology 2.0. These virtual machines are suitable for scenarios like batch processing, web servers, analytics and gaming.
For persistent storage, use the variant Fs virtual machines and purchase Premium Storage separately. The pricing and billing meters for Fs sizes are the same as F-series.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| F1 | 1 | 2.00 GiB | 16 GiB | $- | $- | $- | |
| F2 | 2 | 4.00 GiB | 32 GiB | $- | $- | $- | |
| F4 | 4 | 8.00 GiB | 64 GiB | $- | $- | $- | |
| F8 | 8 | 16.00 GiB | 128 GiB | $- | $- | $- | |
| F16 | 16 | 32.00 GiB | 256 GiB | $- | $- | $- |
Fs-Series
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| Fs1 | 1 | 2.00 GiB | 4 GiB | $- | $- | $- | |
| Fs2 | 2 | 4.00 GiB | 8 GiB | $- | $- | $- | |
| Fs4 | 4 | 8.00 GiB | 16 GiB | $- | $- | $- | |
| Fs8 | 8 | 16.00 GiB | 32 GiB | $- | $- | $- | |
| Fs16 | 16 | 32.00 GiB | 64 GiB | $- | $- | $- |
Memory optimised
High memory-to-core ratio. Great for relational database servers, medium to large caches and in-memory analytics.
E2-64 v3 latest generation
E2-64 v3 instances are the latest generation of hyper-threaded memory optimised instances. E2-64 v3 instances are based on the 2.3 GHz Intel XEON® E5-2673 v4 (Broadwell) processor and can achieve 3.5 GHz with Intel Turbo Boost Technology 2.0. E2-64 v3 instances are ideal for memory-intensive enterprise applications.
Persistent storage disks are billed separately from virtual machines. To use premium storage disks, use the variant Esv3 virtual machines. The pricing and billing meters for Esv3 sizes are the same as Ev3-series. See pricing for disks.
| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| E2 v3 | 2 | 16.00 GiB | 50 GiB | $- | $- | $- | |
| E4 v3 | 4 | 32.00 GiB | 100 GiB | $- | $- | $- | |
| E8 v3 | 8 | 64.00 GiB | 200 GiB | $- | $- | $- | |
| E16 v3 | 16 | 128.00 GiB | 400 GiB | $- | $- | $- | |
| E20 v3 | 20 | 160.00 GiB | 500 GiB | $- | Not available | Not available | |
| E32 v3 | 32 | 256.00 GiB | 800 GiB | $- | $- | $- | |
| E64i v3 1 | 64 | 432.00 GiB | 1,600 GiB | $- | $- | Not available | |
| E64 v3 | 64 | 432.00 GiB | 1,600 GiB | $- | $- | $- |
E2s-64s v3 latest generation
ESv3-series instances are based on the 2.3 GHz Intel XEON ® E5-2673 v4 (Broadwell) processor and can achieve 3.5GHz with Intel Turbo Boost Technology 2.0 and use premium storage. Ev3-series instances are ideal for memory-intensive enterprise applications.
Persistent storage disks are billed separately from virtual machines. To use premium storage disks, use the variant Esv3 virtual machines. The pricing and billing meters for Esv3 sizes are the same as Ev3-series. See pricing for disks.
| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| E2s v3 | 2 | 16.00 GiB | 32 GiB | $- | $- | $- | |
| E4s v3 | 4 | 32.00 GiB | 64 GiB | $- | $- | $- | |
| E8s v3 | 8 | 64.00 GiB | 128 GiB | $- | $- | $- | |
| E16s v3 | 16 | 128.00 GiB | 256 GiB | $- | $- | $- | |
| E20s v3 | 20 | 160.00 GiB | 320 GiB | $- | Not available | Not available | |
| E32s v3 | 32 | 256.00 GiB | 512 GiB | $- | $- | $- | |
| E64is v3 1 | 64 | 432.00 GiB | 864 GiB | $- | $- | Not available | |
| E64s v3 | 64 | 432.00 GiB | 864 GiB | $- | $- | $- |
D11-15 v2
D11-15 v2 instances are based on the 2.4 GHz Intel Xeon® E5-2673 v3 (Haswell) processor and can achieve 3.1 GHz with Intel Turbo Boost Technology 2.0. D11-15 v2 are ideal for memory-intensive enterprise applications. D15 v2 instance is isolated to hardware dedicated to a single customer.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| D11 v2 | 2 | 14.00 GiB | 100 GiB | $- | $- | $- | |
| D12 v2 | 4 | 28.00 GiB | 200 GiB | $- | $- | $- | |
| D13 v2 | 8 | 56.00 GiB | 400 GiB | $- | $- | $- | |
| D14 v2 | 16 | 112.00 GiB | 800 GiB | $- | $- | $- | |
| D15 v2 | 20 | 140.00 GiB | 1,000 GiB | $- | $- | $- |
DS11-S15 v2
For persistent storage, use the variant Dsv2 virtual machines and purchase Premium Storage separately. The pricing and billing meters for Dsv2 sizes are the same as Dv2-series.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| DS11 v2 | 2 | 14.00 GiB | 28 GiB | $- | $- | $- | |
| DS12 v2 | 4 | 28.00 GiB | 56 GiB | $- | $- | $- | |
| DS13 v2 | 8 | 56.00 GiB | 112 GiB | $- | $- | $- | |
| DS14 v2 | 16 | 112.00 GiB | 224 GiB | $- | $- | $- | |
| DS15 v2 | 20 | 140.00 GiB | 280 GiB | $- | $- | $- |
G-series
G-series virtual machines feature Intel® Xeon® processor E5 v3 family and provide unparalleled computational performance to support large database workloads, specifically SAP HANA, SQL Server, Hadoop, DataZen, and Hortonworks. G5 instance is isolated to hardware dedicated to a single customer.
If you need an option to run SAP HANA production environments that require nodes larger than 0.5 TB, the SAP HANA on Azure (large instances) provide purpose-built SAP certified hardware to deploy larger production environments of SAP workloads (OLTP and OLAP). See more details about SAP HANA on Azure.
For persistent storage, use the variant GS virtual machines and purchase Premium Storage separately. The pricing and billing meters for GS sizes are the same as G-series.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| G1 | 2 | 28.00 GiB | 384 GiB | $- | – – | – – | |
| G2 | 4 | 56.00 GiB | 768 GiB | $- | – – | – – | |
| G3 | 8 | 112.00 GiB | 1,536 GiB | $- | – – | – – | |
| G4 | 16 | 224.00 GiB | 3,072 GiB | $- | – – | – – | |
| G5 | 32 | 448.00 GiB | 6,144 GiB | $- | – – | – – |
Gs-Series
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| Gs1 | 2 | 28.00 GiB | 56 GiB | $- | – – | – – | |
| Gs2 | 4 | 56.00 GiB | 112 GiB | $- | – – | – – | |
| Gs3 | 8 | 112.00 GiB | 224 GiB | $- | – – | – – | |
| Gs4 | 16 | 224.00 GiB | 448 GiB | $- | – – | – – | |
| Gs5 | 32 | 448.00 GiB | 896 GiB | $- | – – | – – |
M-series
M-series virtual machines are hyper-threaded and feature Intel® Xeon® E7-8890 v3 2.5GHz (Haswell) processor. The M-series provides up to 128 cores and 3.8 TiB of memory, providing unparalleled computational performance to support large in-memory workloads.
All the M-series virtual machine sizes will have the ability to use both standard and premium persistent disks. The pricing and billing meters for Ms sizes are the same as M-series.| Add to estimate |
Instance | vCPU | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| M32ls | 32 | 256.00 GiB | 1,000 GiB | $- | $- | $- | |
| M64ls | 64 | 512.00 GiB | 2,000 GiB | $- | $- | $- | |
| M8ms | 8 | 219.00 GiB | 250 GiB | $- | $- | $- | |
| M16ms | 16 | 438.00 GiB | 500 GiB | $- | $- | $- | |
| M32ms | 32 | 875.00 GiB | 1,000 GiB | $- | $- | $- | |
| M64ms | 64 | 1,792.00 GiB | 2,000 GiB | $- | $- | $- | |
| M128ms | 128 | 3,892.00 GiB | 4,000 GiB | $- | $- | $- | |
| M64s | 64 | 1,024.00 GiB | 2,000 GiB | $- | $- | $- | |
| M128s | 128 | 2,048.00 GiB | 4,000 GiB | $- | $- | $- | |
| M32ts | 32 | 192.00 GiB | 1,000 GiB | $- | $- | $- |
SAP HANA on Azure Large Instances
Get purpose-built SAP HANA infrastructure with high-performance compute, storage and network supporting up to 24TB RAM on a single node. SAP HANA on Azure Large Instances provide cutting-edge Intel Xeon E7-8890 V4 processors coupled with high-performance NFS storage and networking and provide built-in support for backups through storage snapshots, high availability and disaster recovery and scale-out configurations.
Please contact your Microsoft account team to get pricing for larger SKUs which provide up to 24TB RAM on single node.
| Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|
| S96 | 96 | 768.00 GiB | 3,072 GiB | – – | $- | $- |
| S192 | 192 | 2,048.00 GiB | 8,192 GiB | – – | $- | $- |
| S192m | 192 | 4,096.00 GiB | 16,384 GiB | – – | $- | $- |
| S192xm | 192 | 6,144.00 GiB | 16,384 GiB | – – | $- | $- |
| S384 | 384 | 4,096.00 GiB | 16,384 GiB | – – | Not available | $- |
| S384m | 384 | 6,144.00 GiB | 18,432 GiB | – – | Not available | $- |
| S384xm | 384 | 8,192.00 GiB | 22,528 GiB | – – | Not available | $- |
Extreme Memory Optimised
These new VM sizes are optimised for database workloads like SQL Server or Oracle which often require high memory, storage and I/O bandwidth, but not a high CPU core count. They constrain the VM vCPU count to reduce the cost of software licensing, all while maintaining the same memory, storage and I/O bandwidth. The licensing charged for SQL Server or Oracle will be constrained to the active vCPU count while the compute cost, which includes OS licensing, remains the same one as the original size based on “underlying vCPU”. These new VM sizes which are only available in Azure, allowing workloads to push higher CPU utilisation at a fraction of the (per-core) licensing cost.
| Add to estimate |
Instance | Active vCPU / Underlying vCPU |
RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| DS11-1 v2 | 1 / 2 | 14.00 GiB | 28 GiB | $- | – – | – – | |
| DS12-1 v2 | 1 / 4 | 28.00 GiB | 56 GiB | $- | – – | – – | |
| DS12-2 v2 | 2 / 4 | 28.00 GiB | 56 GiB | $- | – – | – – | |
| DS13-2 v2 | 2 / 8 | 56.00 GiB | 112 GiB | $- | – – | – – | |
| DS13-4 v2 | 4 / 8 | 56.00 GiB | 112 GiB | $- | – – | – – | |
| DS14-4 v2 | 4 / 16 | 112.00 GiB | 224 GiB | $- | – – | – – | |
| DS14-8 v2 | 8 / 16 | 112.00 GiB | 224 GiB | $- | – – | – – | |
| E4-2s v3 | 2 / 4 | 32.00 GiB | 64 GiB | $- | – – | – – | |
| E8-2s v3 | 2 / 8 | 64.00 GiB | 128 GiB | $- | – – | – – | |
| E8-2s v3 | 4 / 8 | 64.00 GiB | 128 GiB | $- | – – | – – | |
| E16-4s v3 | 4 / 16 | 128.00 GiB | 256 GiB | $- | – – | – – | |
| E16-8s v3 | 8 / 16 | 128.00 GiB | 256 GiB | $- | – – | – – | |
| E32-8s v3 | 8 / 32 | 256.00 GiB | 512 GiB | $- | – – | – – | |
| E32-16s v3 | 16 / 32 | 256.00 GiB | 512 GiB | $- | – – | – – | |
| E64-16s v3 | 16 / 64 | 432.00 GiB | 864 GiB | $- | – – | – – | |
| E64-32s v3 | 32 / 64 | 432.00 GiB | 864 GiB | $- | – – | – – | |
| Gs4-4 | 4 / 16 | 224.00 GiB | 3,072 GiB | $- | – – | – – | |
| Gs4-8 | 8 / 16 | 224.00 GiB | 3,072 GiB | $- | – – | – – | |
| Gs5-8 | 8 / 32 | 448.00 GiB | 6,144 GiB | $- | – – | – – | |
| Gs5-16 | 16 / 32 | 448.00 GiB | 6,144 GiB | $- | – – | – – | |
| M64-16ms | 16 / 64 | 1,750.00 GiB | 2,000 GiB | $- | – – | – – | |
| M64-32ms | 32 / 64 | 1,750.00 GiB | 2,000 GiB | $- | – – | – – | |
| M128-32ms | 32 / 128 | 3,800.00 GiB | 4,000 GiB | $- | – – | – – | |
| M128-64ms | 64 / 128 | 3,800.00 GiB | 4,000 GiB | $- | – – | – – |
Storage optimised
High disk throughput and IO. Ideal for Big Data, SQL and NoSQL databases.
L-series
L-series instances are storage optimised virtual machines for low latency workloads such as NoSQL databases (e.g. Cassandra, MongoDB, Cloudera and Redis). L-series offer up to 32 CPU cores, using the Intel® Xeon® processor E5 v3 family with 8-GiB random-access memory (RAM) per core and from 768 GB to 6 TB of local solid state drive (SSD) disk. This instance provides Premium Storage support.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| L4 | 4 | 32.00 GiB | 678 GiB | $- | $- | $- | |
| L8 | 8 | 64.00 GiB | 1,388 GiB | $- | $- | $- | |
| L16 | 16 | 128.00 GiB | 2,807 GiB | $- | $- | $- | |
| L32 | 32 | 256.00 GiB | 5,630 GiB | $- | $- | $- |
GPU
Specialised virtual machines targeted for heavy graphic rendering and video editing available with single or multiple GPUs.
NC-series
N-series virtual machines are ideal for compute and graphics-intensive workloads, helping customers to fuel innovation through scenarios like high-end remote visualization, deep learning and predictive analytics. N-series virtual machines feature the NVIDIA Tesla accelerated platform and NVIDIA GRID 2.0 technology, providing the highest-end graphics support available in the cloud today. In addition, N-series offers a NC24r configuration that provides a low latency, high-throughput network interface optimised for tightly coupled parallel computing workloads.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| NC6 | 6 | 56.00 GiB | 340 GiB | 1X K80 | $- | $- | $- | |
| NC12 | 12 | 112.00 GiB | 680 GiB | 2X K80 | $- | $- | $- | |
| NC24 | 24 | 224.00 GiB | 1,440 GiB | 4X K80 | $- | $- | $- | |
| NC24r | 24 | 224.00 GiB | 1,440 GiB | 4X K80 | $- | $- | $- |
NCv2 series
NCv2 series virtual machines is a new addition to the GPU product family offering the next generation of our popular NC-series machines, powered by NVIDIA Tesla P100 GPUs. Customers can take advantage of these updated GPUs for traditional HPC workloads that will benefit from a performance boost, powering scenarios like reservoir modeling, DNA sequencing, protein analysis, Monte Carlo simulations and others. These new GPUs can provide more than 2 times the computational performance of the current NC-series. And like our NC series, we offer a configuration with InfiniBand networking for workloads that require fast interconnect, like Oil and Gas, Automotive and Genomics to also accelerate scaleout capability as well as improved single instance performance. In addition, N-series offers an NC24r configuration that provides a low latency, high-throughput network interface optimised for tightly coupled parallel computing workloads.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| NC6 v2 | 6 | 112.00 GiB | 336 GiB | 1X P100 | $- | $- | $- | |
| NC12 v2 | 12 | 224.00 GiB | 672 GiB | 2X P100 | $- | $- | $- | |
| NC24 v2 | 24 | 448.00 GiB | 1,344 GiB | 4X P100 | $- | $- | $- | |
| NC24r v2 | 24 | 448.00 GiB | 1,344 GiB | 4X P100 | $- | $- | $- |
NCv3 series
NCv3 series virtual machines is a new addition to the GPU product family offering the next generation of our popular NC-series machines, powered by NVIDIA Tesla V100 GPUs. Customers can take advantage of these updated GPUs for traditional HPC workloads that will benefit from a performance boost, powering scenarios like reservoir modeling, DNA sequencing, protein analysis, Monte Carlo simulations and others. These new GPUs can provide 1.5 times the computational performance of the current NCv2-series. And like our other NC series, we offer a configuration, NC24r v3, with InfiniBand networking for workloads that require fast interconnect, like Oil and Gas, Automotive and Genomics to also accelerate scale out capability as well as improved single instance performance.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| NC6 v3 | 6 | 112.00 GiB | 336 GiB | 1X V100 | $- | $- | $- | |
| NC12 v3 | 12 | 224.00 GiB | 672 GiB | 2X V100 | $- | $- | $- | |
| NC24 v3 | 24 | 448.00 GiB | 1,344 GiB | 4X V100 | $- | $- | $- | |
| NC24r v3 | 24 | 448.00 GiB | 1,344 GiB | 4X V100 | $- | $- | $- |
NV-series
N-series virtual machines are ideal for compute and graphics-intensive workloads, helping customers to fuel innovation through scenarios like high-end remote visualization, deep learning and predictive analytics. N-series virtual machines feature the NVIDIA Tesla accelerated platform and NVIDIA GRID 2.0 technology, providing the highest-end graphics support available in the cloud today.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| NV6 | 6 | 56.00 GiB | 340 GiB | 1X M60 | $- | $- | $- | |
| NV12 | 12 | 112.00 GiB | 680 GiB | 2X M60 | $- | $- | $- | |
| NV24 | 24 | 224.00 GiB | 1,440 GiB | 4X M60 | $- | $- | $- |
NVv2-series preview
NVv2 series virtual machines are powered by Nvidia Test M60 GPUs and NVIDIA GRID technology with Intel Broadwell CPUs. These virtual machines are targeted for GPU accelerated graphics applications and virtual desktops where customers want to visualise their data, simulate results to view, work on CAD or render and stream content. Additionally, these virtual machines can run single precision workloads such as encoding and rendering. NV_v2 virtual machines support Premium Storage and come with twice the system memory (RAM) when compared with its predecessor NV Series.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| NV6s v2 | 6 | 112.00 GiB | 736 GiB | 1X M60 | $- | – – | – – | |
| NV12s v2 | 12 | 224.00 GiB | 1,474 GiB | 2X M60 | $- | – – | – – | |
| NV24s v2 | 24 | 448.00 GiB | 2,948 GiB | 4X M60 | $- | – – | – – |
ND-series
ND-series virtual machines is a new addition to the GPU family specifically designed for the needs of the AI and deep learning workloads offering excellent performance for training and inference. It is powered by NVIDIA Tesla P40 GPUs based on the new Pascal Architecture. These instances provide excellent performance for FP32 (single precision floating point operations), for AI workloads utilising CNTK, TensorFlow, Caffe and other frameworks. The ND-series also offers a much larger GPU memory size (24 GB), enabling it to fit much larger neural net models. Like our NC-series, the ND-series offers a configuration with a secondary low-latency, high-throughput network through RDMA and InfiniBand connectivity so you can run large-scale training jobs spanning many GPUs. In addition, N-series offers an ND24r configuration that provides a low latency, high-throughput network interface optimised for tightly coupled parallel computing workloads.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
GPU |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|---|
| ND6 | 6 | 112.00 GiB | 336 GiB | 1X P40 | $- | $- | $- | |
| ND12 | 12 | 224.00 GiB | 672 GiB | 2X P40 | $- | $- | $- | |
| ND24 | 24 | 448.00 GiB | 1,344 GiB | 4X P40 | $- | $- | $- | |
| ND24r | 24 | 448.00 GiB | 1,344 GiB | 4X P40 | $- | $- | $- |
High performance compute
Our fastest and most powerful CPU virtual machines with optional high-throughput network interfaces (RDMA).
H-series
The H-series is a new family specifically designed to handle high performance computing workloads such as financial risk modeling, seismic and reservoir simulation, molecular modeling and genomic research. They are based on the Intel Xeon E5-2667 v3 Haswell 3.2 GHz (3.6 GHz with turbo) with DDR 4 memory. Turbo is always on for H-series instances. Two of the H-series configurations (H16r, H16mr) also carry a second low latency, high-throughput network interface (RDMA) optimised and tuned for tightly coupled parallel computing workloads such as MPI applications.
| Add to estimate |
Instance | Core | RAM | Temporary storage |
Pay as you go
|
One year reserved (% Savings) |
Three year reserved (% Savings) |
|---|---|---|---|---|---|---|---|
| H8 | 8 | 56.00 GiB | 1,000 GiB | $- | $- | $- | |
| H16 | 16 | 112.00 GiB | 2,000 GiB | $- | $- | $- | |
| H8m | 8 | 112.00 GiB | 1,000 GiB | $- | $- | $- | |
| H16m | 16 | 224.00 GiB | 2,000 GiB | $- | $- | $- | |
| H16mr | 16 | 224.00 GiB | 2,000 GiB | $- | $- | $- | |
| H16r | 16 | 112.00 GiB | 2,000 GiB | $- | $- | $- |
Linux support
Support for certain Linux virtual machine images in the Azure Marketplace is available from the publisher. In some cases, it is provided by the publisher for free. In other cases, we charge by the minute for the right to receive support. The total charge for running a Linux virtual machine is the support rate (if applicable) plus the Linux compute rate. Microsoft does not provide support for Linux images in the Azure Marketplace.
Free trial, MSDN, BizSpark and Microsoft Partner Network member benefits (monetary credits) do not apply to publisher provided Linux support, which is billed separately.
Canonical Ubuntu Server Linux images
Customers who deploy virtual machines from Canonical Ubuntu Server Linux images available in the Azure Marketplace are entitled to receive patching from Canonical for those virtual machines at no additional charge over the compute rates.
OpenLogic CentOS-Based Linux Images
Customers who deploy virtual machines from OpenLogic CentOS-based Linux images available in the Azure Marketplace will be entitled to receive patching from OpenLogic for those virtual machines at no additional charge over the compute rates.
IP address options
Every Azure Cloud service containing one or more Azure Virtual Machines is automatically assigned a free dynamic virtual IP (VIP) address. For an additional charge, you can also get:
- Instance-level public IP addresses—A dynamic public IP address (PIP) that is assigned to a virtual machine for direct access.
- Reserved IP addresses—A public IP address that can be reserved for your subscription, which can be used as a VIP address for any cloud service in the region.
- Load-balanced IP addresses—Additional load-balanced VIP addresses that can be assigned to an Azure Cloud Service containing one or more Azure Virtual Machines.
Support and SLA
- Free billing and subscription management support.
- Flexible support plans starting at $29/month (Note—Microsoft support does not include support for Linux images in the Azure Marketplace, though in some cases Linux publishers offer support). Find a plan.
- Guaranteed 99.95 percent connectivity for multiple instances. Read the SLA.
FAQ
-
No. All new virtual machines have an operating system disk and a local disk (or “resource disk”). We do not charge for local disk storage. The operating system disk is charged at the regular rate for disks. See all virtual machines configurations.
-
Read our Azure Reserved Virtual Machine Instances FAQ.
-
A temporary disk is storage that is attached directly to the deployed VM. Data on a temporary disk is lost once the VM is shut down. If you need persistent storage, there are a number of types and sizes of persistent data disks available that are charged separately. Please see storage pricing for information. For persistent VM storage, we recommend that you use Managed Disks to take advantage of better management features, scalability, availability and security.
-
Maybe. If the status says “Stopped (Deallocated),” you are not being billed. If it says, “Stopped Allocated,” you are still being billed for allocated virtual cores (not the software licence itself). Full details on virtual machine states are available here.
State Billed Details Starting Yes The initial starting state of virtual machines as they are going through the boot cycle. This period is billed as the virtual machines are running. Running (Started) Yes The running state of the virtual machine. Stopped Yes You are billed for allocated cores, but not software licence.
Note—To put a virtual machine in “Stopped” status, use the power option within the virtual machine, not from the Microsoft Azure classic portal.Deleted (Deallocated) No Cores are no longer allocated to the virtual machine and are no longer billed.
Note—The only way to stop all virtual machines in a cloud service is to delete both the staging and production deployments. -
Virtual machine prices include local disk storage only. You can see the cost to attach Premium (SSD based) and Standard (HDD based) disks to your virtual machines on the Managed Disks pricing page.
-
Previously, all usage for A0, A2, A3 and A4 Standard and Basic virtual machines was emitted as fractions (for A0) or multiples (for A2, A3 and A4) of A1 virtual machines meter minutes. We heard this caused some confusion for our customers, so we have implemented a change to assign per-second usage to dedicated A0, A2, A3 and A4 meters. Because of this transition, the resource GUIDs for A0, A2, A3 and A4 virtual machines will change. If any of your billing routines rely on the resource GUIDs, they will need to be modified to take these new billing meters into consideration.
Deployment size Usage emitted as multiple of A1 up to 10th January 2016 Usage emitted on dedicated meter as of 11th January 2016 A0 .25 of A1 hour 1 of A0 hour A2 2 of A1 hour 1 of A2 hour A3 4 of A1 hour 1 of A3 hour A4 8 of A1 hour 1 of A4 hour -
We charge for the number of full minutes your virtual machine is running, so you are not billed for any extra seconds. In this example you would be billed for 6 minutes.
-
It’s a new preview feature for Azure Virtual Machines with private IP addresses. The internally load-balanced endpoint is only accessible within a virtual network (for virtual machines within a virtual network) or cloud service (for virtual machines outside a virtual network). Internal load balancing is useful for multi-tier applications where some of the application tiers aren’t public facing yet require load balancing functionality.
-
You can check the status of your virtual machines in the Virtual Machines tab and also in the virtual machine Dashboard. If virtual machine status says "Stopped," you are still being billed. Only "Stopped (Deallocated)" indicates you are not being billed.
To ensure you are not being billed, always stop virtual machines from the management portal. You can also stop the virtual machine through Powershell by calling ShutdownRoleOperation with "PostShutdownAction" equal to "StoppedDeallocated".
-
Each instance is charged for its data egress across regions.
-
No. Tax will be added separately.
-
Yes, through licence mobility, if you have Software Assurance you can "bring-your-own-licence" for all Virtual Machines supported server products.