Used for ex nightly batch processing at banks. Lots of horse power for your on prem needs. Moving this kind of horse power to the cloud would be insanely expensive and complex for the mid size banks relying on these systems.
They shipped these massive boxes out and connected them to massive SANs. You could license the processors later if your workload grew.
I spent a year as consultant to big customers building SOA, websites, SSO for whatever they needed. ATM networks with low latency etc.
They came with avcertain number of power cores, say 8 but you could start by only licensing 4 and then grow into them. Don't remember the exact specs on this but they were crazy powerful 20 years ago.
The mainframe is System Z. The power line is meant to replace things like Windows Servers and EC2 instances.
You would typically install software like your CRM, ERP and web servers on the IBM Power systems. These would then talk to the mainframe (System Z / Telum) to handle any extremely high stakes business activity.
A healthy all-in IBM organization would be using both of these technologies for what they're best suited for. If you run salesforce and your GH enterprise instance on the mainframe, you are going to be spending a LOT of money compared to the alternative.
Can confirm that 'cromulent' is in the Oxford English Dictionary [0] ...
>> Acceptable, adequate, satisfactory.
>> Frequently used humorously or ironically in recognition of its origin as an invented word in the television programme The Simpsons (see quot. 1996).
As is 'embiggen', namely 'transitive. To make bigger or greater, to enlarge.'
What they were providing sounded fairly niche in the Chipsandcheese interview. Terabytes of DRAM on an entirely different protocol than DDR (less "stressed" with robust interconnects and a higher signal integrity - which I would assume necessarily comes with a higher cost for boards and silicon real estate), strongly parallel and specifically designed to improve signal quality/reroute around bad connections, and generally hyper focus on uptime for mission critical massive in memory databases.
I've never understood why these processors really exist before, but I think that makes sense.
The traditional Z mainframes (focused on uptime to the point where everything is hot swappable while running and redundant) I did understand as probably having some valid use deep inside the financial system and defense, but the enterprise facing solutions like power I never really got.
Anyway if anyone has more to add I'd like to hear it. Is my first paragraph mostly it?
AIX on big iron.
Used for ex nightly batch processing at banks. Lots of horse power for your on prem needs. Moving this kind of horse power to the cloud would be insanely expensive and complex for the mid size banks relying on these systems.
They shipped these massive boxes out and connected them to massive SANs. You could license the processors later if your workload grew.
I spent a year as consultant to big customers building SOA, websites, SSO for whatever they needed. ATM networks with low latency etc.
They came with avcertain number of power cores, say 8 but you could start by only licensing 4 and then grow into them. Don't remember the exact specs on this but they were crazy powerful 20 years ago.
Just starting into a project to move an AIX behemoth to Linux, with clouds (still far) ahead. Fingers crossed...
A noble spirit embiggens the smallest man.
better coverage at https://chipsandcheese.com/p/ibm-power-whats-next?utm_source...
I'm genuinely surprised, I thought the release of the Telum chip signed the death of the POWER processors.
Are they meant to be two different tiers of mainframe processors?
The mainframe is System Z. The power line is meant to replace things like Windows Servers and EC2 instances.
You would typically install software like your CRM, ERP and web servers on the IBM Power systems. These would then talk to the mainframe (System Z / Telum) to handle any extremely high stakes business activity.
A healthy all-in IBM organization would be using both of these technologies for what they're best suited for. If you run salesforce and your GH enterprise instance on the mainframe, you are going to be spending a LOT of money compared to the alternative.
Embiggen is a perfectly cromulent word!
True
And it's a shame they used it incorrectly in the title - "embiggen" means "enlarge, make bigger", not "become bigger" which is the case with Power11
ive just read thats actually become a word in dictionaries now
Can confirm that 'cromulent' is in the Oxford English Dictionary [0] ...
>> Acceptable, adequate, satisfactory.
>> Frequently used humorously or ironically in recognition of its origin as an invented word in the television programme The Simpsons (see quot. 1996).
As is 'embiggen', namely 'transitive. To make bigger or greater, to enlarge.'
[0] https://www.oed.com/dictionary/cromulent_adj?tab=meaning_and...
"...the utter dependence that customers...have on these boxes..." is less technical than contractual.
What they were providing sounded fairly niche in the Chipsandcheese interview. Terabytes of DRAM on an entirely different protocol than DDR (less "stressed" with robust interconnects and a higher signal integrity - which I would assume necessarily comes with a higher cost for boards and silicon real estate), strongly parallel and specifically designed to improve signal quality/reroute around bad connections, and generally hyper focus on uptime for mission critical massive in memory databases.
I've never understood why these processors really exist before, but I think that makes sense.
The traditional Z mainframes (focused on uptime to the point where everything is hot swappable while running and redundant) I did understand as probably having some valid use deep inside the financial system and defense, but the enterprise facing solutions like power I never really got.
Anyway if anyone has more to add I'd like to hear it. Is my first paragraph mostly it?