Oracle customers clamor for its hardware. Yup, hardware. It can't build Exadata fast enough Oracle this week reported two per cent revenue growth for the first quarter of its 2021 financial year, along with strong demand for its hardware that caused a backlog of orders for Exadata and other equipment. Revenue reached $9.367bn, up from $9.218bn in the same quarter last year and over $150m ahead of guidance. Cloud …
Timesharing is dead, get your own mainframe.
Mainframes are expensive and inflexible, move to distributed mini/personal computers to solve all your problems.
Minis/PCs are hard to manage because they're distributed, move to centrally-managed servers & thin clients.
Servers are expensive and inflexible, move to cloud.
Oh, cloud looks just like timesharing.
Rinse & repeat.
I've lived through one complete cycle round the loop, time to retire before we do it all over again.
Well said - I've said the same here, probably not as well.
I DID manage to retire awhile back.
From this perspective, it's (mostly) amusing.
Always remember how much is supported by the churn - which will always encourage further churn - those ad bucks, PR paychecks, even online tech sites. It might even produce employment that is noticeable in the same scale as the tech itself! And talking about stuff is generally less risky than actually doing it.
Yup. And the amusing thing is that every generation hails one of those as the latest greatest and they all think its brand new and novel and has never been seen before on this planet.
I'll give it 5 to 10 years before the next generation of IT admins and C suites get fed up with paying someone else money to rent-a-box and relying on their call centres for support then it'll be back to the company server room once again.
It’s the same in programming languages and software development. I’ve been around long enough now for Tony Hoare’s CSP to have been fashionable twice. Google’s Protocol Buffers is simply a reinvention of things like ASN.1, minus the sophistication. Where once we used makefiles, we now have the equally messy cmake, IDEs having been and gone. JavaScript is worse than BASIC. Web technologies deliver worse results than XServer. About the only proper innovation is Rust, which really does deserve plaudits for having taken the lid off the question, “What is a programming language?”, and answered it properly this time.
I do not disagree with the sentiment, Nursing A Semi, but marketing BS aside there may be other considerations in the case of Exadata. 40Gbps QDR Infiniband storage network with sub-microsecond application-to-application latency would be quite difficult to provide in "the cloud", so if your requirements include that...
[Hiding as AC for the following disclaimer: I was involved in development of the original Exadata networking - not at Oracle, at a vendor... Add salt, pepper, spices to taste before believing anything.]
Exadata is backlogged because :
a) The fed cant print dollar bills fast enough or in high enough denominations to pay what it costs.
b) I have it on good authority that Oracle are offering such massive discounts to at least 1 Tier Infrastructure player such that its cannibalising Oracles direct sales. I suspect thats where the demand spike is coming from, I also suspect they are last in the queue to get hold of the kit.
I will never understand why companies buy HW from Oracle. Even heavily discounted, it's still just x86 servers with tons of Oracle (read: expensive) spaghetti code. This is a few years old, and heavily IBM oriented, but I don't believe they are lying. And I think still relevant: https://www.ibm.com/downloads/cas/WOZDNBYX
Oh good grief, you've reminded me of my time writing code for Oracle DB using C++ and embedded SQL. The Oracle header files were quite easily the worst C++ code I've ever seen, requiring hack upon hack and ignoring screens worth of compiler warnings to get a vaguely working application.
At one contractor site, they got so sick of Oracle and its lock-in licencing model, they trashed EVERY Oracle machine onsite (i.e. all hardware and disks wiped and literally taken to the metal recycler!) and went with an IBM Z-series mainframe setup and IBM supported SQL databases!
$15 Million dollars later they are completely happy with how little downtime they have had (i.e. SIX SIGMA or 99.99966% uptime most years!) and how EASY it is to update and replace gear and that IBM's Hot Site services let you keep going EVEN IF there was a hurricane caused flood at a data site!
Maybe I am really biased towards a BIG IRON style of hardware, but I REALLY LIKE using IBM mainframes even in the Year 2020 for massive online databases and realtime low-latency transactions processing!
For all its faults, IBM STILL MAKES AWESOME mainframe-class client/server hardware!
V
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
