Updated #2: How Elastic is the Oracle Cloud

Per-Second billing now on almost all resources!

June 2020 Update:
– Database Cloud Service uses per second billing for VM, Baremetal and Exadata
– Updated Exadata CS for the new X8 Shapes
– Added list of OCI Autoscaling features, including the new schedule based scaling for instance pools

May 2020 update:
– Compute now uses per second billing
– Autonomous Database now uses per second billing
– New E3 Compute shape that supports flexible shape in the form of # of CPUs

March 2020 update:
– Compute VMs now support automated shape change
– Database VMs now support automated shape change
– Exadata Shapes now only have a minimum term of 48 hours (instead of 30 days)

One of the key characteristics of Cloud Computing is elasticity; resources that are available on-demand when needed and stopped when not needed, resources that can scale to the current demand. This elasticity is not only great to ensure well performing applications, but can also significantly help to reduce the cost of such solutions.

Let’s take an Example: You want to have an Analytics Environment to discover valuable data for your business. An analytics environment typically only needs to run when the user is actually working with it. This will clearly not be 24/7. The database behind the analytics environment might also only need to run during the use of analytics. So in some cases both can be powered off when not needed, resulting in significant cost saving.

Service	Cost 24×7 per Month	Cost 10×5 per Month	Savings
Analytics Standard @ 2 OCPU	$2,000	$896	$1,104 (55.2%)
Autonomous Data Warehouse @ 2 OPCU	$2,500	$874	$1,626 (65.1%)
Total Monthly cost	$4,500	$1,769	$2,731 (60,7%)

In case you are wondering why the savings on Autonomous DB is different then on Analytics cloud; Autonomous has zero CPU cost when not running, while the Analytics cloud charges 15% of CPU while not running.

Save Money, but get POWER when needed!

In the above example you see the savings these services can give you if you run them “smartly”, but sometimes you just want to go fast! These service support scaling so you will have that option. The above configuration/price is based on 2 OCPUs for each service but:

• Autonomous DB can scale WITHOUT downtime from 1 to 128 OCPUS
• AnalyticsCS can scale WITHOUT downtime between 2,4, 6 or 8 OCPUS

So if you need some extra juice, you can! You will only be paying extra for the duration (in hours) for the extra capacity.

Oracle Cloud Services Elasticity Overview

Service	Suspend	Suspend Cost	Scale Up/Down	OCPU Range	Scale In/Out	Comments
Compute VM	Yes	Storage Cost	Change shape requires a reboot	Intel and E2 Shapes: 1, 2, 4, 8, 16, 24 E3 Shapes: Any CPU between 1 and 64	via Instance Pool	Stopping of billing only for Standard shapes not based on Windows OS. Per-Second billing
Compute BM	Yes	Storage cost	No	36,44,52,64, 128	No	Stopping of billing only for Standard shapes not based on Windows OS, Per-Second billing (min one hour)
Compute Instance Pool	Yes	Storage Cost	Change shape requires a reboot		Yes, 1 to many nodes	Per-Second billing
Database VM	Yes	Storage Cost	Change shape requires a reboot, when using RAC no downtime needed	1, 2, 4, 8, 16, 24	Yes, RAC 2 Nodes	Per second billing
Database Bare Metal	No		Yes	2 – 52 OCPU	No	DB license for OCPU is scalable. There is a fixed infrastucture cost for the BM server. Per second billing
Exadata CS Quarter X8	No		Yes	0 – 100 OCPU	RAC 2 Nodes	Fixed Infra cost. Minimum term is 48 hours, there after per hour. Per second billing
Exadata CS Half X8	No		Yes	0 – 200 OCPU	RAC 4 Nodes	Fixed Infra cost. Minimum term is 48 hours, there after per hour. Per second billing
Exadata CS Full X8	No		Yes	0 – 400 OCPU	RAC 8 Nodes	Fixed Infra cost. Minimum term is 48 hours, there after per hour. Per second billing
Autonomous DB	Yes	Storage Cost	Yes	1 – 128 OCPU	Above 16 OCPU split on multiple RAC Nodes	Per-Second billing
Analytics CS	Yes	15% of OCPU hour rate	Yes	(2,4,6,8) or (10,12)	No	Shapes 1, 16, 24,236,52 do not provide scaling
Java CS	Yes	Storage Cost	No		Yes, 1,2 or 4 Nodes
Visual Builder CS	Yes		No		No
Big Data Cloud	Yes	Storage Cost	No	2,4,8,16,24	Yes, 1 to 100 nodes

Native OCI Auto-scaling Features

OCI is getting more and more native auto scaling features. Currently there are 2 services that have an auto scaling method.

• Instance Pools; Based on CPU or Memory load, Schedule based auto-scaling
• Autonomous DB; Automatic CPU scaling based on load up to 3x the default size.

Auto-scaling by Scripting

If you want to scale other things besides Instance Pools or the CPU of an autonomous database, you can my Auto Scale Script

The Super Scale” script that can automatically power on/off and scale Compute and Database Instances. I will also soon be adding support for the Analytics cloud and some other services in this script, so “To be continued…” 🙂

https://github.com/AnykeyNL/OCI-AutoScale