Unbreakable Availabiilty Domain ...
... go on ... someone say it! Ha ha ha.
Nope ..."miles apart".
Oracle will open a cloud region in the UK that's based in London and plans to have three data centres serving the region by mid-2017. The region will accompany others in Virginia and Turkey in a plan Oracle stated will double "the regional presence of its cloud platform in the last 24 months, with 29 regions available globally …
Each region will consist of three high-bandwidth, low-latency sites, named Availability Domains, which will be based within several miles of each other and operate fault-independently. It's unclear whether Oracle will build or lease the facilites required to float the UK portion of its cloud.
Is anyone else worried by the lack of insight of the part in bold.
What with the speed of light being as it is and fibre being plentiful, there is no reason why they can't increase "several miles" to something more aligned with recognised best practice on designing resilient infrastructure.
Understanding the market should be fairly high up if you want to compete with the big boys in the market.
General practice is that adjacent availability zones have low latency, mostly in-order packet delivery and low likelihood of a network partition.
If you want higher levels of isolation then go multi region and design your architecture to copy with network partitions.
Seems like a waste of time and cash. Oracle can play in SaaS, application software, but IaaS and PaaS... no way they can compete with AWS and Google's volumes/pricing and technology stacks. Oracle's issue, like MSFT to a lesser extent, is that Oracle wants to make legacy margins on Exadata and DB and WebLogic to please Wall St. It just isn't going to work when AWS and Google are going to sell the infrastructure service for an incredibly low price and give away the DB and middleware software for nothing.
There's no reason Oracle can't join in the IaaS game. They're way behind but they've got pots of money to hire ex Google or AWS people, and they are ahead in the application space if they want to do SaaS.
Whether the margins would be as high as they are used to is questionable, but the alternative might be to die out.
At the moment there's no PaaS Exadata-like service. If there were I'd be spending a lot on it. AWS's SPICE looks like it might evolve if something but that's years away.
It is probably possible for Oracle to join the IaaS game if they wanted to spend all of their cash getting their infrastructure on par with AWS and Google, but there is no ROI in it... spend tens of billions on data centers and fiber runs all over the world, only to have to beat AWS and Google on price.... when AWS and Google's cloud services are subsidized by their huge volumes of consumer internet services. That is kind of key to being a cloud provider for IaaS and PaaS... hyper scale consumer services to subsidize the massive infrastructure and software costs.
"no way they can compete with AWS and Google's volumes/pricing and technology stacks."
Lol, Google? They are not even in the picture versus Oracle and hardly anyone is using them for Pass / Iaas. The real competitor here for Oracle is Microsoft who are wiping the floor with Oracle in terms of SQL server and Azure...
This post has been deleted by its author