2020年4月2日木曜日

LINUX OS running on a mainframe (general purpose machine) in a 19" rack

Click here for Japanese Version.

Click here for FLEXCUBE search results



Related Information:

https://neovisionconsulting.blogspot.com/2019/07/blog-post_1.html?m=1

COMMENT:

All over the world, WEB servers for banks and credit cards are being used in a variety of ways, from old mainframes (general-purpose machines) to

On LINUX on the latest mainframe made by IBM that fits in the rack mount of 19 inches (general-purpose machine), such as FLEXCUBE and other accounting systems, and as of 2019, in the above environment, Java WEB sites and intranet seems to have been introduced a lot. The Cyclone framework for C10K is currently undergoing maintenance for the upgrade from Python2 to Python3, and may be compatible with AI and machine learning. I hope to see more AI and machine learning API liveliers in the future.



On the combination of an excellent ideal programming language and a framework that Masahiro Ishizuka thinks
1, Support for C10K (10,000 simultaneous client access problem)
thing.

2、They have AI and machine learning libraries.

3, in the development efficiency in the team, code visibility and in a good sense

It is non-individual and cannot express variety in a good way.

It's something that you can get a sense of.

4. Security (vulnerability) issues are unlikely to occur.

5、Multi-core and multi-thread performance.

6、The program must be easy to read and write.

7、The callback hell has been taken care of.

8、It must be a compilation language.

9, No concept of a pointer.

10, cryptographic library for around electronic signatures
In Java and Go, the encryption library for the electronic signature seems to be implemented, but other languages are unconfirmed.
I know only Go lang+Gin and Python+Cyclone are the most likely ones for 1 to 3 above. But there are still many frameworks for the Go language.
Nim is the next fastest language after C, but it seems to be able to express a lot of variety, so I don't know if I should call it non-individualistic. That and the fact that we don't know at this point if there is a machine learning library.
Julia doesn't seem to be as fast as I thought she would be, so it's disappointing.
The Go language and machine learning libraries are summarized below.
http://neovisionconsulting.blogspot.com/2018/08/gotensorflow-prediction.html?m=1
The advantage of running a website developed in Go language is that it uses less memory.
Python has the most extensive AI and machine learning libraries, and the ease of development in these areas is an advantage. However, it seems that Go is faster than Python in terms of programming language alone.
Go lang+Gin and Python+Cyclone are both very fast because they both address the 10,000 concurrent client access problem, and can handle 10,000 pieces of data per second, while the Python+Japronto framework apparently supports a million pieces of data per second.


The ideal programming language for all of the above 1 to 5 and 8 is
It will be Rust.
Rust seems to run at about three times the speed of JAVA and Go, so it will become more and more important in mission-critical sites such as banks and credit cards.
As for AI (machine learning) of 2, as of 2020.02.22, Tensorflow binding is open to the public, but it doesn't seem to support GPU.
How to create a sense of unity and integration between teams when developing with Rust.
https://neovisionconsulting.blogspot.com/2020/02/rustwebgolang_11.html

Nim is the one that supports 1 to 9 above, but all but 3 and 4.
Nim's multi-core, multi-threaded performance is under investigation.
Nim seems to be the next fastest in the benchmarking process after the optimized C language.
Nim has also addressed the C10K issue.
The advantage and disadvantage is that Nim can be programmed with a Python like syntax.
The programming writing is Python like, so it's easy to read and write, visibility is good, and development efficiency in a team is not bad.
However, compared to Golang, it can be more expressive in a bad way, so if you don't have a solid coding rule in your team, if you develop a website or an intranet without much discussion, it's hard to get a sense of unity.
If Nim had a strict formatter like Python or Go, it would be easier to create a unified team and solve the above three problems.
It's a lot of work to fix because we both got different things than we expected.
The best way to ensure a consistent quality is to use a user-defined pramma at compile time to check for errors, warnings, hints, etc.
https://blog.hirokiky.org/entry/2019/06/03/202745
Reference information.
(I won't argue about how to write Python in detail anymore, I'll just auto-format it in black.)
I hope that Nim will be able to add an option to perform a security check for weaknesses, including memory leaks, strictly at compile time, like Rust, or that this will be improved in future versions.
Fastest < Optimized C < Nim < Rust < Go < JavaScript < Slowest > The left side of the screen seems to be faster.
http://h-miyako.hatenablog.com/entry/2015/01/23/060000
FYI: I shared.
...


(The Nim version is the fastest for Pandas, Rust is the best overall power for websites!



https://neovisionconsulting.blogspot.com/2020/03/the-nim-version-is-fastest-for-pandas.html


...
As of 2020.02.02.
In Python 2 & 3, the global interpreter lock problem has not been solved.

As an alternative method.
I think the only way to do this is to use another true multi-core, true multi-threaded programming language.
-----------.
Rust has a bit of a difficult concept of pointers and ownership, but it is safe, multi-core, multi-threaded, and clears vulnerability issues with benefits that outweigh its slightly more difficult disadvantages as it learns and clears them, while benchmarking results are third to Nim after optimized C, three times faster than Go or Java, and fast enough to withstand mission-critical tasks such as processing credit card systems with worldwide access on LINUX on a mainframe (general purpose machine) in a 19" rack, and bank accounts with worldwide access.
-----------.
Become familiar with Rust's ownership.
https://qiita.com/cactaceae/items/2c70a9947364c60ec100
ooooh!
Why we moved from Python to Go
https://frasco.io/why-we-switched-from-python-to-go-19581e27de7c.
Shared.
...
Benchmarking with popular languages (Dart, Go, Julia, Nim, Python, Rust, etc.)
http://h-miyako.hatenablog.com/entry/2015/01/23/060000
ooooh!
Goodbye Python, hello Go!
https://yakst.com/ja/posts/5253.
Shared.
----
Changed from Go to Rust to improve performance
https://www.atmarkit.co.jp/ait/articles/2002/10/news038.html

Spanner (Google) DATABASE
Click here for Google search results
https://www.google.com/search?sxsrf=ALeKk02AMOqDbwLwyqk5NyM89j-yZAGK2Q%3A1583816172122&source=hp&ei=7B1nXoOkBdGNoASBgrr4CA&q=Spanner%EF%BC%88Google%EF%BC%89&oq=Spanner%EF%BC%88Google%EF%BC%89&gs_l=psy-ab.3..0i30l7j0i8i30.2912.2912..6486.... .2.0..0.82.82.82.1....................................................................................................... .0.... .2j1...gws-wiz.... .10...35i362i39.wemX0KXGBw4&ved=0ahUKEwiDys6Cj4_oAhXRBogKHQGBDo8Q4dUDCAg&uact=5
NewSQL DATABASE with high speed ACID
CockroachDB official version Ver1.0 is released.
The official version of CockroachDB, aimed at the Google Spanner clone, Ver. 1.0 was released.
VoltDB
What is Open Source VoltDB?
It also meets the requirements of FoundationDB and ACID. VoltDB, a SQL database that surpasses NoSQL, benchmarked against Cassandra!
It also meets the requirements of ACID, a NoSQL database FoundationDB that writes 14 million times per second. VoltDB, a SQL database that surpasses NoSQL, benchmarked against Cassandra!


 
     
11
     
37
     
 2
     
     
notification
Koichi Tanigawa[writer]

Edited by DB Online 2018/05/24 06:00



 Lately, I've been writing a lot less about hardware. At the Dell Technologies World in the United States during Golden Week, the PowerEdge MX, a modular server that has been redesigned to keep up with the latest technology for a longer period of time, was announced. Also at IBM THINK in March, a collaboration to make NVIDIA's GPUs more efficient on POWER 9 processor-based machines was discussed, which was a bit of a refreshing thought.






































IBM adopts IBM Z for its blockchain services platform

0 コメント:

コメントを投稿