– Richard Phillips
In the rapidly evolving world of cloud technology, the question of which service to adopt is generally seen as a choice between Microsoft Azure and AWS with Google Cloud computing services bringing up the rear and IBM and Oracle some way behind [1]. However, for large-scale online transaction processing, Oracle still leads the way as the world’s most popular database management system [2] and Oracle’s cloud based offering deserves serious attention.
If we consider a typical enterprise there is likely to be a sizable existing investment in Oracle database technology. When moving services to the cloud, however, AWS or Azure have quickly become the go-to providers. Substantial investment in Oracle is effectively written off in favour of spend on new capability. To combat this rush to AWS and Azure, Oracle launched their Autonomous Database in 2018 but it’s still not seen as the first choice for many enterprises; even those with large existing Oracle investment. The reasoning behind this may be a perception that Oracle is expensive, complicated and arduous to deal with. Oracle’s introduction of its autonomous database on its own cloud is an attempt to address this perception and has gone quite some way to bringing a challenger to AWS and Azure.
The Autonomous Database quite simply runs itself without downtime: it does all the administration, optimisation, patching, memory management and, to some extent, scaling, automatically. This is an incredibly significant move for cloud providers. Cloud databases have evolved significantly but, for the most part, you still need technical teams to administer and keep them running at their optimal level. Microsoft has managed some of this with their Azure SQL database but elements such as optimisation of indexes are still down to the administrator and the database is not as full featured as Redshift or Oracle. Oracle has removed the burden completely by using AI to supplement their in-house teams keeping the system running as part of the service. As an example of how much they have simplified things, the admin manual for standard Oracle 19c database runs to a total of 1690 pages (about the same as the first two Game of Thrones books) and that’s just the admin section – this is almost entirely taken care of in the autonomous version of the same database. With most of the admin taken away, the team of expensive Oracle DBAs are free to do something else (maybe retrain in Amazon Redshift which still has a high demand for admins).
The next thing to consider is pricing. As is the case for most cloud database platforms the price is based around compute and storage with discounts based on committed usage. Following this model the price comes out around 10% more expensive than the equivalent service using AWS. But if we then factor in the lower administrative burden, the total cost of ownership comes out strongly in favour of Oracle. We need to remember, however, that many enterprises have already made a strong financial commitment to their on premise Oracle licences and it is here that the biggest advantage becomes clear. The licences are fully portable to the autonomous version and if you follow this “bring your own” licence model the cost for the service is going to be around 1/3 of the cost of an equivalent AWS.
As far as performance goes the database wins hands down in the published benchmark reports against AWS Redshift and even the same version of Oracle running within the AWS RDS platform. Performance is between 4 and 14 times as fast than the equivalent workloads [3]. If however more power is needed then scaling the database is done online with the ability to auto scale up and down with zero downtime. This extra compute can be used to deploy “just in time” extra resource when needed automatically and scale back again when not to save costs. Unlike other platforms there is no admin overhead as the system manages scaling based on the query demand and can automatically scale up to three times the original compute levels. Importantly you only pay when the extra compute is being used.
The final point to make is around availability. We normally talk here of 99.9% or 99.95% guaranteed availability, but Oracle have gone quite a bit further here offering 99.995% availability. The most important part of this guarantee is what it means for updates. When standard maintenance is your responsibility, then the downtime associated with this is not included in the availability – you must take the hit yourself. For Redshift on AWS you need to schedule in a 30 minute maintenance window every week. With SQL Azure there is no window so your connection will drop at the time of update and you just need to make sure your code allows for this. As the admin is included in the Oracle offering, the uptime includes all the updates and they happen automatically with no outage. For mission critical services this may well be the most important factor.
The database is of course just one part of your cloud strategy and it’s quite widely accepted that the rest of Oracle’s cloud offering is still evolving. For now an all-in approach with Oracle is probably not a sensible approach but many would argue that a multi cloud approach spanning more than one provider is a more sensible approach for many enterprises. The risk is spread and you retain leverage over the providers when it comes to renegotiation of contracts. As it stands, for transactional and analytical needs the Oracle Autonomous Database should almost certainly form a large part of any serious multi cloud strategy.
[1] Source RightScale 2019 State of the cloud report from Flexera
[2] Source Gartner June 2019
[3] Source Oracle OpenWorld tests October 2018
