Translated from 日本語 • View original Hello. This is Hashimoto, an infrastructure engineer.
Like saeki, we are participating in re:Invent 2017, a major event that AWS is proud of.
Check-in error even before departure, and on the first day, I had a hard time moving by bus (for some reason it is not in the order described in the pamphlet&
When I arrived, everyone got off without saying a word), but I got used to it, so I worked hard to report on the second day.
Here is the state of the first day.
Global Partner Summit
Digest Key Contents
- This year, the adoption of AWS is progressing, especially in the healthcare and financial sectors, such as cardiovascular care and improving the efficiency of banking systems.
- Expand the partner ecosystem (a mechanism that flexibly provides cloud value from among various partner companies to meet the business needs of end users) and give the impression that AWS is absolute.
- We will continue to strengthen the IOT, ML, Blockchain, and AI fields
- Enterprises are gradually migrating to AWS. Currently, it is the United States, but the next trend will be global
- There is no option to not move (not migrate) from there (on-premise) for databases. For example, a certain oracle. You have to choose which side you will be on (the venue is buzzing)
- 65 new Premier Partners
Tuesday Night Live
Keynote by Peter Desantis, VP of Global Infrastructure at AWS.
About Regions
11 additional regions will be added in 2016-2018. Government Region to the United States. The first region in the Middle East (Bahrain) also appeared.
Starting with Sweden, new regions will be launched using renewable energy
About AvailabilityZone
- An explanation of the basic origin, such as having more than 100,000 servers in multiple DCs with physical distance.
- AZ has a minimum of 2 zones, but some regions have up to 6 zones
- 100 GB of Amazon Global Network connected regions and connected to AWS with 67 DirectConnect
Computing at SCALE
Machine Lerning (hereinafter referred to as ML)
Deep learning and X-ray are actually used in the medical field
For example: X-ray examination to find brain tumors, etc.
Why we have been able to grow so far
(1) Hardware sophistication
②ML framework improvement
AWS **Apache MXNet**Machine learning training using was 7-8 times more effective
③Rapidly growing
- GLUON
- Open neural network exchange
- NNVM and TVM
Whether it's using machine learning model APIs or switching tools that train on one engine and then run it in another.
It seems to have a phenomenal learning effect.
These tools seem to be driving the dramatic speed of machine learning.
acceleration for all framework on all hardware
Accelerate all frameworks on all hardware
This was the message.
Bare Metal Instance Announcement
- If there is something that cannot be virtualized
- If you need a special HyperVisor
- When using a special host license
It is suitable for the above cases.
Announcing Amazon Guard Duty
A new security service that detects threats in the AWS environment.Simply enable it to provide IDS/IPS/SIEM.
Agentless, no need to think about log sizing in advance.
It can also be used in the Tokyo region.
Conclusion
On the first and second days, I don't have enough time to move around, including sessions.
In terms of presentations, I felt that the fields of IOT, AI, and machine learning have continued to be exciting, and by combining them with security and automation, we are realizing new services and things that existing services could not do before.
After all, the time when automation will no longer require human hands is approaching.
It's past 3 o'clock at the time of writing this article, but when I wake up, Keynote is waiting for me! It's been a long time since I've had such a childlike excitement.
Next time, I would like to update the content mainly in Keynote.