I listened to an Oracle webinar on “The Value of Engineered Systems” yesterday. It didn’t really cover anything ground-breaking from my point of view, but it offered the following facts or thinking:
It should be noted that I’ve spent my career as a Oracle (relational) DBA of some description, so I’m sure there’s some degree of bias. However, I’m an advocate of Oracle databases because they’ve proven to be the best on the market.
- There will be an estimated 50 billion Internet devices by 2020.
- 90% of all data in the world has been created within the last 2 years.
- An estimated 50x data growth is expected by 2020 thanks to the Internet of Things, Social Media, surveillance
- An estimated 340 trillion IP addresses will be required by 2020 – I hope SOMEONE is using IPv6 by then!
- IT departments need to grow by 4% per annum just to maintain existing systems, excluding any growth of business needs – yet companies continue to make cuts in infrastructure and avoid investing in R&D.
- Storage accounts for 17% of IT budget.
- Oracle do NOT expect customers to host their PRODUCTION systems in the PUBLIC Cloud – they see customers to host PRODUCTION in a PRIVATE Cloud (either hosted by the customer or via co-lo) ***
- Instead, they expect SaaS for Development to be the big draw for the PUBLIC Cloud.
- Engineered systems simplifies IT.
- Oracle is awesome.
- Exadata is awesome on steroids.
I was surprised to hear that Oracle do not expect customers to host their Production systems in the Public Cloud, despite the latest trends in the industry media. Instead, they see customers hosting their Production systems in a Private Cloud, either in their own data center or at a co-location facility.
I wasn’t sure whether this was a case of “old” thinking simply not being updated for the webinar, but I did check with some trusted Oracle types who confirmed that this IS Oracle’s current thinking.
From a DBA/DMA point of view, this makes sense – as it probably does from a security administrator’s as well. But, if you were to buy into the hype, you’d think that Oracle was a year or two out of date and that they hadn’t fully embraced the reality of what’s happening on the ground in 2014.
It remains to be seen who’s right. Maybe both sides will. Certainly, without major changes in technology, I have strong doubts that cloud hosting could match/supplant the performance of the engineered systems such as Exadata; specialist needs require specialist, not standardized, solutions – and just MOVING your multi-terabyte enterprise data warehouse into the PUBLIC cloud on commodity hardware is only the first of many challenges.
If you believe the hype, Oracle is a dead man walking. It’s interesting to note that so many trade mag “articles” extolling the virtues of the Next Big Thing(TM) always include a variation of “Oracle-killer” in their byline. Whereas the standard relational database model is not necessarily suited for EVERY data management task (such as HUGE unstructured data and “Big Data”), that’s not to say that engineered systems – or, more precisely, an integrated suite of engineered system comprised of different disciplines, including relational databases (with an in-memory capability?), “NoSQL” databases and standard “Big Data” tools – should be excluded from the gamut of possible solutions.
Exadata, for instance, can be tuned to run OLTP loads and, using Exadata Hybrid Columnar Compression, Writable Flash Cache and Flash Cache Compression, you could keep a large part – if not all – of your database in memory.
Is it scalable? Yes, it is. It’s expensive and difficult to do, but so is “rolling your own” alternative. All that happens is that you swap “hard” costs such as a hardware for “soft” costs such as customization of a particular environment.
Are engineered systems “cloudifiable”? Aren’t “cloud” and “engineered systems” mutually exclusive? I can’t see engineered systems making it to the cloud, unless Oracle do it, though they could be hosted at a co-location facility.
That way, the hardware spend doesn’t go to waste and a lot of the potential problems with cloud hosting don’t apply. They also have the significant benefit of having end-to-end professional support for both hardware and software and being a known, mature and supportable quantity – by DBAs and developers alike.
Oracle isn’t going anywhere – nor is Exadata – any time soon. And the cloud will have to make major leaps forward before it’s able to be offer a realistic alternative to the features and performance that Oracle databases, and Exadata machines specifically, provide.
If you need a Ferrari to handle your business workloads, it doesn’t matter how cheap a Kia Soul is, nor whether it’s in the cloud or not.