Oracle Analytics Cloud Performance

As part of a recent POC I was working on I’ve done some tests on #OracleAnalyticsCloud performance vs an AWS VM and an on-premises instance – lets see how fast OACS is.

In each case, I had 2x fairly large cubes that I tested calc performance against. The cubes were identical, containing approx 100million datapoints (including a fair few zeros).
I tested 3 different servers to get a feel for how OAC performs compared to essbase that we’ve all used for years. I also tested fairly large cubes to check that there were no obvious limits or blockers to shifting large cubes onto the cloud.

The Servers

On Premises

This is a multi-purpose machine I use for lots of demos, testing & internal systems – so it’s pretty busy.

  • 4x Core
  • 16Gb
  • Running lots of apps (Apex, DB, SVN, Wiki, Apps Dev, Essbase, EAS, OBIEE, VA, Publisher). Running >80% Memory Utilisation.


I was kindly granted temporary access to a VM version of OACS – it’s the one you get access to when you run a demo via I added a little more horsepower to it since I wanted to run the RDBMS and the BI stack as well to test out some other functions:

  • 8x Core
  • 32Gb
  • Running full OAC Stack (DB, Apex, Essbase, BI, DVCS)

OAC on Oracle

This was running the a “full” #OracleAnalyticsCloud setup. I used the minimum specification I could. I didn’t start up a BI node at all – for this test, I didn’t need it. So we were running a RDBMS node and a seperate Essbase node only.

  • 1x OCPU Essbase
  • 1x OCPU RDBMS (2 cores per OCPU, btw)
  • 8Gb

The Cubes

I have tested ASO calc performance on 2x POC cubes for one of our customers.

  • Both cubes are ASO Unicode.
  • Neither are particularly optimised.
  • Approx 100 million data points loaded
  • All versions across the servers are identical

The Test

I ran a simple MAXL aggregation and timed the results:
execute aggregate process on database 'ABI_Reg'.'ABI_Reg' stopping when total_size exceeds 10;

You can see I set the aggregation factor to be quite high – these cubes have a number of flat dimensions, so making it do a bit more pre-aggregration helps enomrmously.

The Results

  • We’d expect the AWS OAC VM to be the fastest (it has more cores) – and probably the OAC & on premises servers to be similar (the on-premises server is pretty busy, even though it has more cores).
  • Performance is roughly as we expected – OAC is in the ballpark of what we usually see.

The Learning

  • With Oracle, 1x OCPU is 2x Cores.
  • When sizing an OAC Essbase Node, use the same established on-premises sizing guidelines to ensure you have enough CPU & Memory to calculate the cubes. Particularly relevant if there are lots of users, or big cubes that need calculating.
  • The other learning point is that these cubes need tuning. 🙂

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s