- Oracle presents the Oracle Cloud as if it is a competitive cloud offering.
- In addition to functionality shortcomings, Oracle Cloud has location shortcomings as well.
It all started with a conversation. The customer was angry. I listened, asked a few questions, and decided to investigate. You can read the customer story in his own words here.
The problem is repeated outages with Oracle cloud. Why so many and why do they last so long? Also, the standard practice with AWS, GCP, and Azure is that outages are compensated with service credits. Why are Oracle’s cloud outages different? Why don’t they come with cloud credits!
We decided to investigate and answer three questions:
- Why does Oracle cloud have an unusually high number of outages?
- Why do outages last so long?
- Why customers don’t receive service credits for these outages?
Let us go over each item.
Why Does Oracle Cloud Have an Unusually High Number of Outages?
Oracle cloud is offered in 5 locations: Phoenix, Ashburn, London, Frankfurt, and most recently, Toronto.
Are these data centers owned or operated by Oracle?
The answer is NO.
These are all leased colocation facilities owned and operated by Digital Reality and Cologix. You can read the announcements by these colocation operators in the links provided. Oracle also uses Equinix and other interconnectivity providers. Here’s the announcement from Digital Reality:
And the recent announcement for the Toronto facility from Cologix:
Oracle cloud is entirely made up of leased colocation facilities operated by third-party providers. These third party providers have their own maintenance schedules for systems maintenance, updates, and upgrades. They also experience their own frequent unplanned downtime. Third party downtime adds additional downtime to Oracle’s own planned and unplanned downtimes. Customers have to deal with downtime from Oracle PLUS downtime from third-party colocation providers. This explains the Oracle Cloud’s high frequency of outages.
Why do Outages Last So Long?
When something breaks, there’s no single throat to choke. Oracle support blames the colocation provider and the provider turns around and blames Oracle. The provider often blames an ISP or the interconnect operator. Too many moving parts and no single ownership. This makes problem resolution extremely difficult and time-consuming. Keep in mind that these third-party providers don’t have a private fiber network connecting their facilities to other provider facilities. All communication has to go through the public internet. With so many cross-connects with backbone and regional networks, troubleshooting problems get exponentially harder and take much longer.
Why Customers Don’t Receive Cloud Credits for Outages?
There are two reasons.
First, Oracle cloud SLA doesn’t recognize outages caused by third parties as eligible for service credit. This applies to metered and un-metered services as you can read on page 5, section 3.3, The Definition of Unplanned Downtime.
When downtime, planned or unplanned, is caused by a third party provider, customers get no service credits.
Secondly, Oracle does NOT credit customers for Oracle’s own planned maintenance downtime. It actually says that in the Oracle cloud policy!
This means that downtime caused by Oracle maintenance, updates, and upgrades either for Oracle OCI or customer specific environments doesn’t count toward any customer service credits.
The problem with Oracle’s cloud is Oracle’s low level of commitment to the cloud. This is clear from Oracle’s anemic CAPEX spend on cloud infrastructure compared to other providers like AWS, Microsoft, and Google.