The silver lining to cloud prices (AWS and Azure)

The silver lining to cloud prices (AWS and Azure)

AWS and Azure are currently the leading cloud service providers.  Unfortunately, their pricing structures can be difficult to understand. We have summarised their product offerings below.  In short, Amazon offers more flexible pricing models while Azure’s pricing is often simpler , providing more control of the underlying technology.

azure and aws

Amazon Web Services

Amazon offers a myriad of product and services within Amazon Web Services.  The most commonly used service being Elastic Cloud Computing and Simple Storage Service. AWS services operate out of geographic regions, with pricing varying between regions.

EC2

EC2 (Elastic Cloud Computing) is AWS’ virtual machine infrastructure capable of hosting operating systems including Windows and Linux on a myriad of system configurations. For general purpose needs the hardware ranges from 1 GiB of RAM with 1 CPU on “t2.micro” up to 32 GiB of RAM with 8 CPUs on “t2.2xlarge”. Other instance types are available with hardware optimised for specific needs such as the Accelerated Computing (p2) instances with Graphics Processing Units for heavy parallel processing of data (such as machine learning or 3D graphics rendering for special effects).
EC2 instances are charged via 4 different pricing models, each optimised for a specific use case.

On-Demand (Pay only for uptime)

The On-Demand model is the most flexible in that you will only pay for storage and VM uptime. If your application is only needed a few months of the year, it can stay shut down for example.

Reserved Instance (Like a bulk purchase)

With a reserved instance you pay for the whole term’s usage up front, but actually, save from 20% to 50% compared to the on-demand price for the same usage. This model is good if you have consistent and constant usage. You still pay the same cost even if the instance is shut down for some of the time however. Reserved Instances can also be applied to Availability Zones with the capacity shared across multiple instances.

Spot Instance (Bid on spare capacity)

For some use cases, paying the cheapest rate is more important than immediate availability, for example, a demonstration or experiment. In these cases, you could bid on the spare capacity of other EC2 computing availability, with savings of up to 90% of the On-Demand price.

Dedicated Hosts (Your own dedicated physical machine)

Using a Dedicated Host can help save you money by reusing existing software licences; address corporate and regulatory compliance with the knowledge it’s running on a physical server you control. Once you’ve chosen your dedicated host’s size, are charged using either the on-demand or reserved instance model, though at a higher rate.

Sample Pricing (Asia Pacific Region – Sydney)

(Note that pricing may vary depending on the enterprise agreement.)
t2.micro Instance (1 virtual CPU, 1GB RAM, EBS storage)

Operating System

Price

Pricing Model

Linux

$0.016 per hour

On-Demand

Linux

$103 per year ($0.012 /hour)

Reserved 1 Year Term All Upfront

Windows

$0.020 per hour

On-Demand

Windows

$140 per year ($0.016 /hour)

1 year term All Upfront

m3.medium Instance (1 virtual CPU, 3.75GB RAM, 1×4 SSD)

Operating System

Price

Pricing Model

Linux

$0.093 per Hour

On-Demand

Linux

$481 per year ($0.055 /hour)

Reserved 1 Year Term All Upfront

Linux

$0.0122 per Hour

Spot Instances

Windows

$0.020 per hour

On-Demand

Windows

$140 per year ($0.016 /hour)

Reserved 1 year term All Upfront

Windows

$0.0721 per Hour

Spot Instances

$3.274 per Hour

Dedicated Host On-Demand

$16955 per year ($1.936 /hour)

Dedicated Reversed

For more information see: https://aws.amazon.com/ec2/pricing/

S3

S3 – Simple Storage Service – is Amazon’s object storage model for general purpose data storage. It is commonly used for user content, shared data, or web assets. S3 is available in 3 different classes optimised for access time. Data stored in S3 is charged by the amount stored per month and becomes cheaper per GB with high volumes. As well as data storage, there are also fees for each request, though these are quite small.

Standard

Standard is s general purpose storage class with high availability and frequent access. S3 Standard is suitable for static web sites content, uploaded content or data being manipulated or output from some business function.

Infrequent Access Storage

The same availability as standard but with less frequent write access. Intended for objects that aren’t updated as often, this model provides a storage discount over Standard though with higher access fees. (But still quite small)

Glacier

Finally, Glacier is Amazon’s archival object storage class, well suited for storing backups on. It is considerably cheap to store data but is intended for long term storage. As such the data has a minimum storage time of 90 days, and in fact has minimum access time delays, though for a fee the request for access can be expedited.

Following is a sample of S3 pricing and availability options (in Sydney Region):
(Note that pricing may vary depending on the enterprise agreement.)

cloud pricing

cloud pricing

Azure

Microsoft’s cloud platform, Azure, also offers many different products and services for cloud computing needs, with an object store and virtual machine infrastructure being two of the more common.

Block Blob storage

Block Blobs are Microsoft Azure’s object storage service for storing and streaming documents, videos, images, backups and other unstructured data in a similar vein as Amazon’s S3. Azure services are also divided by Region (the region of the world where the data centre is located) and the storage and transfer pricing may vary.
Azure Block Blob’s main pricing model is based on data redundancy rather than access time. Within each redundancy option, storage tiered pricing also applies, giving discounts for high volume customers. (Redundancy in data storage parlance refers to having multiple copies of the same data on different hardware or data centres for increased reliability.) The data redundancy options are:

  • Locally Redundant Storage (LRS) – mirrors data within the one data centre. This is the cheapest option, though a physical data centre level disaster may cause the data to be lost.
  • Zone Redundant Storage (ZRS) – mirrors data across multiple data centres within the same or multiple regions. This
  • Geographically Redundant Storage (GRS) – provides multiple mirrors within the primary data centre, plus mirrors the data to a secondary data centre a safe distance away.
  • Read-access geo-redundant storage (RA-GRS) – same as GRS, but provides direct read-only access to the secondary data centre. This is also the default option upon creation of a storage account.

The Azure Block Blob’s storage model provides much more visibility into the data redundancy available than does Amazon’s S3, but S3 may be more cost effective depending on region and zone. For more information see: https://azure.microsoft.com/en-au/services/storage/blobs/

Azure Virtual Machines

Azure Virtual Machines are available in various usage categories – General Purpose, Compute/Memory/Storage Optimised, VMs optimised for graphics processing and a new "H Series" optimised for high-performance computing such as molecular modelling and genomic research or seismic and reservoir simulation. Azure only has one primary pricing model available and is charged per minute which may be more cost effective for VMs with only small uptimes. Azure also has a spare capacity tier available called Azure Batch which can offer significant savings over standard, high priority VMs. Following is a sample of some VMs and their prices.
(Note that pricing may vary depending on the enterprise agreement.)

INSTANCE

CORES

RAM

DISK SIZES

PRICE

LOW PRIORITY PRICE

A1 v2

1

2.00 GiB

10 GB

$0.058/hr

$0.009/hr

D1 v2

1

3.50 GiB

50 GB

$0.101/hr

$0.015/hr

D1

1

3.50 GiB

50 GB

$0.125/hr

$0.017/hr

N.B. from Azure website: Storage values for disk sizes use a legacy “GB” label. They are actually calculated in gibibytes, and all values should be read as “X GiB”.
For detailed pricing see https://azure.microsoft.com/en-au/pricing/details/batch/ and https://azure.microsoft.com/en-au/pricing/details/virtual-machines/linux/

Canopy Tools

CT4 are actively developing a new product called Canopy Tools for simplified cloud and on-premises VM maintenance and reporting, initially supporting Azure and VM Ware. In its initial release Canopy Tools will provide a unified view of an organisation’s virtual machines across VM Ware and Azure – providing monitoring and logging of disk, RAM, and CPU usage over time; running costs of cloud VMs and cost comparison of on-premises VMs. As well as a view of VMs, Canopy Tools will also support instance maintenance such as VM shutdown and deployment of a given specification. In future releases, we will add support for more cloud providers, more reporting and deeper configuration and management.
We are experts in cloud-based and on-premises virtual machine maintenance. Contact us today for further information.



Want to keep up to date with the latest industry news and products? Subscribe here and never miss a beat.