EXT4 Gets A Nice Batch Of Fixes For Linux 5.8

[kloczek]

https://www.sustainable-bus.com/news...ied-this-year/

https://www.statista.com/statistics/...ue-since-2005/

[S.Pam]

No mention of Solaris at all.

Why are we feeding this troll? https://tse2.mm.bing.net/th?id=OGC.a...mh%2bRJc1QQ%3d

[Volta]

https://www.statista.com/statistics/...ue-since-2005/

Oracle's revenue, but in unreadable form. I have something better:

https://www.macrotrends.net/stocks/charts/ORCL/oracle/revenue

Oracle Corporation is a global provider of enterprise cloud computing and is empowering businesses of all sizes on their journey of digital transformation. Oracle Cloud provides leading-edge capabilities in Software as a Service (SaaS), Platform as a Service (PaaS), Infrastructure as a Service (IaaS), and Data as a Service (DaaS). Oracle's security solutions enable organizations to implement and manage consistent security policies across the hybrid data center. Oracle security cloud services make leading security technologies available everywhere to organizations large and small. Oracle Cloud Platform enables developers, IT professionals, and business leaders to develop, extend, connect, and secure cloud applications, share data, and gain insights across applications and devices. Oracle's application suites, platforms, and infrastructure leverage both the latest technologies and emerging ones, including artificial intelligence (AI), machine learning, blockchain, and Internet of Things (IoT).

Too bad there's no even Solaris mentioned, so I'm still waiting.

[Volta]

No mention of Solaris at all.

Why are we feeding this troll?

Yep, there's no point in wasting our time. Trolls used to be better in the past.

[kloczek]

A single Linux kernel supported 4096 CPUs years ago. How many slowlarises you saw? How many cores does Solaris support?

M8 HW can fit up to 16 M8 CPU sockets. Each one is 32 cores and each core is with up 16/8 hardware threads. 16*16*32=8192. It fits in single 48U rack.

You are talking about single system HW on which is used as partitioned system with separated Linux kernel working inside of each of those partitions.
That hardware is used that way because biggest problem is to provide fast enough low latency interconnects between all those systems. If you will look closer on that HW you will find that avg bandwidth used is about 100Kb/s to few Mb/s. All those interconnects are mostly used by MPI horizontally scaled HPC applications.

So .. no this not what I've been talking about

I'm taking about thousands of CPU HW with terabytes of RAM working under SIS (SIS stands for Single Image System).
Most of those systems are used as platform not for HPC but for in-memory databases. With Solaris and compressed ARC on systems with 16-32TB of RAM you may fit in memory database which total size is bigger than total installed size of installed RAM. Oracle DB with columnar compression on some databases provides can provide +10 times compression ratio.
There are some number of big databases which does not scale at all across many smaller physical system and must be installed on single instance HW.
Mostly that kind of databases today are used by telecoms (try for example to think about finding quickly route to the exact mobile phone handset when you are dialling exact number)

[kloczek]

while:

how thin slowlaris market is?

Only few tents billions $ ..

[Volta]

M8 HW can fit up to 16 M8 CPU sockets. Each one is 32 cores and each core is with up 16/8 hardware threads. 16*16*32=8192. It fits in single 48U rack.

Ok, the last feeding. I don't care about hardware capacity, but I asked for Solaris number of CPUs support. Furthermore, it states the M8 scales up to 256 cores.

You are talking about single system HW on which is used as partitioned system with separated Linux kernel working inside of each of those partitions.
That hardware is used that way because biggest problem is to provide fast enough low latency interconnects between all those systems. If you will look closer on that HW you will find that avg bandwidth used is about 100Kb/s to few Mb/s. All those interconnects are mostly used by MPI horizontally scaled HPC applications.

No, I'm talking about the number of CPUs a single Linux kernel supports.

So .. no this not what I've been talking about

It's funny, because I'm feeling the same.

I'm taking about thousands of CPU HW with terabytes of RAM working under SIS (SIS stands for Single Image System).
Most of those systems are used as platform not for HPC but for in-memory databases. With Solaris and compressed ARC on systems with 16-32TB of RAM you may fit in memory database which total size is bigger than total installed size of installed RAM. Oracle DB with columnar compression on some databases provides can provide +10 times compression ratio.
There are some number of big databases which does not scale at all across many smaller physical system and must be installed on single instance HW.
Mostly that kind of databases today are used by telecoms (try for example to think about finding quickly route to the exact mobile phone handset when you are dialling exact number)

And for a single image system there's such support I wrote earlier. Ps. why are you pasting some old thread bullshit?

[Volta]

Only few tents billions $ ..

You're still repeating Oracle market value which is something far different.

[kloczek]

And for a single image system there's such support I wrote earlier. Ps. why are you pasting some old thread bullshit?

OK.
Could you please point on documentation of that system?

[Volta]

OK.
Could you please point on documentation of that system?

Of course:

https://www.kernel.org/doc/html/latest/x86/x86_64/mm.html

https://access.redhat.com/articles/rhel-limits