With the latest ODA X8 range, at least 80% of the customers could find an ODA configuration that fits their needs. For the others, either they can’t afford it, either they are already in the Cloud, or they need extremely large storage or EXADATA performance. Among these 80% of customers, only a few choose ODA. Let’s see how Oracle could improve the ODA to make it a must.
1) Make robust and reliable releases
This is the main point. ODA is built on Linux, Grid Infrastructure and database software, nearly identical to what you can find on a classic linux server. But it comes bundled and with odacli, a central CLI to manage deployment, database creations, updates, migrations and so on. And it sometimes has annoying bugs. More reliable releases could also make patching less tricky, and customers much more confident in this kind of operation.
It could also be nice to have long-term releases on ODA, like on the database. One long-term release each year, with only bug fixes and deeply tested, for those customers who prefer stability and reliability: most of them.
2) Make a real GUI
An appliance is something that eases your life. You unpack the server, you plug it, you press the power button, and you start configuring it with a smart GUI. ODA is not yet that nice. GUI is quite basic, and most of us use the CLI to have a complete control over all the features. So please Oracle, make the GUI a real strength of the ODA, with a pinch of Cloud Control features but without the complexity of Cloud Control. That would be a game-changer.
3) Integrate Data Guard management
Data Guard works fine on ODA, but you’ll have to setup the configuration yourself. Most of the customers plan to use Data Guard if they are using Enterprise Edition. And actually, ODA doesn’t know about Data Guard. You’ll need to configure everything like if it were a standard server. In my dreams, an ODA could be paired up with another one, and standby databases automatically created on the paired ODA, duplicated and synchronized with the primaries. Later we could easily switchover and switchback from the GUI, without any specific knowledge.
There is a lot of work to achieve this, but it could be a killer feature.
4) Get rid of GI for ODA lites
Yes, Grid Infrastructure adds complexity. And a “lite” appliance means simplified appliance. GI is needed mainly because we need ASM redundancy, and ASM is really nice. It’s actually better than RAID. But do you remember how ASM was configured in 10g? Just by deploying a standalone DBhome and creating a pfile with instance_type=ASM. That’s it. No dependencies between ASM and the other DBHomes. Once ASM is on the server, each instance can use it. And it could make patching easier for sure.
5) Make IPs modifiables
Because sometimes you would need to change the public IP address of an ODA, or its name. Moving to another datacenter is a good example. For now, changing IPs is only possible when appliance is not yet deployed, meaning unused. You can eventually change the network configuration manually, but don’t consider future patches will work. An easy function to change the network configuration on a deployed ODA would be welcome.
6) Be proud of this product!
Last but not least. Yes, Cloud is the future. And Oracle Cloud Infrastructure is a great piece of Cloud. But it will take time for customers to migrate to the Cloud. Some of them are even not considering Cloud at all for the moment. They want on-premise solutions. ODA is THE solution that perfectly fits between OCI and EXADATA. It’s a great product, it’s worth the money and it has many years to live. To promote these appliances, maybe Oracle could make ODA better integrated with OCI, as a cross-technology solution. Being able to backup and restore the ODA configuration to the Cloud, to put a standby database in OCI from the GUI, to duplicate a complete environment to the Cloud for testing purpose, …
ODA is a serious product, but it still needs several improvements to amplify its popularity.