Takeaways from the 2019 Paris AWS Summit
Four members of the BCG Platinion Paris team attended the Amazon Web Services (AWS) Summit in Paris on the 2nd of April 2019. In this article, they will share some of their thoughts and insights on this event.
First of all, there was an impressive attendance that could be justified due to the fact that there are no entrance fees, and also free food! Still… Several thousand people, 7000 approximately, and many partners booths made it a significant event in Europe.
A key element during this day was the “x“ factor. Indeed, many new products and features were often confirmed or announced by AWS with this “x” times more/faster than one year ago.
Here are some more details on the announcements that caught our eyes.
1. Constant innovation on databases
First of all, AWS used a very interesting classification of databases.
Then, we witnessed the announcement of two newcomers in the set of offerings at AWS: Timestream and the Quantum Ledger database (QLDB). Amazon Timestream is a fast, scalable, fully managed time series database service for IoT and operational applications that makes it easy to store and analyze trillions of events per day at 1/10th the cost of relational databases.
QLDB is Amazon’s centralized ledger database that comes with all the benefits of an immutable cryptographically verifiable ledger and none of the headaches of having to implement it yourself.
Yes, it is centralized which means it is not distributed trust, but this database is not looking for consensus. Immutability, cryptographically verifiable and serverless are some of the top features that it promotes. Amazon even claims that this centralized, serverless version is 2–3x faster than “common” blockchain. We certainly don’t want common.
At the first sight of this new species of a database in the wild, we thought to ourselves, what could it possibly be used for? Then flashbacks of bygone days when investment banks had to be audit compliant rushed in. Remember the long painstaking hours of rushing to meet and satisfy compliance such as the SEC’s CAT compliance. Knowing that QLDB could make such pains all go away gave a sense utter calm. What ideas can you dream up that QLDB can solve?
2. Leveraging on AI for security
AWS has a defined framework for security, defined in the “AWS well architected” recommendations. One of the tenets of cybersecurity is to “make sure humans remain away from data for more security”. The AWS presentation highlighted the following 3 recommendations: activate traceability, apply security in depth and automate policies
It was very interesting to see the developments of security solutions based on AI. AWS solutions are GuardDuty and Macie. Amazon Guard Duty is a threat detection service that continuously monitors for malicious or unauthorized behavior to protect your AWS accounts and workload. Amazon Macie is a machine learning-powered security service to discover, classify and protect sensitive data.
The orchestration of such tools in an end-to-end security can be illustrated by this drawing:
3. Migrate well and fast to the cloud
This conference was about designing an accelerated path to a successful cloud migration. The testimonial and best practices provided helped us to build a comprehensive approach with the right steps to take, the right data to gather and the right questions to ask.
Migration phases:
The customer’s journey to the cloud adoption typically involves four phases:
- Project: In the project phase, you are running projects to get familiar with the cloud and experience its benefits
- Foundation: After experiencing, you then build the foundation to scale your cloud adoption. This includes creating a landing zone (a pre-configured, secure, multi-account AWS environment)
- Migration: In this stage, you migrate existing applications to the cloud.
- Reinvention: The next step is to focus on reinvention by taking advantage of the flexibility and capabilities of AWS to transform your business by speeding the time to market and increasing the attention on innovation.
Migration Strategies:
They are six Common Application Migration Strategies, Rehost (“lift and shift”), Replatform (“lift, tinker and shift”), Repurchase (“drop and shop”), Refactor / Re-architect, Remove and Retain. 60% of the projects are in ReHost and RePLatform.
Can the migration process be agile?
Yes it can be. The migration process follows four steps:
- Prepare: Determine the objectives of the migration, develop a primary business case for migration that take in account as is architecture.
- Activate : Understand the current applications portfolio and begin to consider migration strategies
- Migrate: Focus on individual application to design, migrate, and validate each application. (discover, build, integrate, validate, feedback)
- Operate: Constantly iterate towards a modern operating model
4. 23+ families of EC2 Instances, but how to choose?
This conference was animated by Arthur Petitpierre, AWS and Stephane Enten, CTO of Adikteev. The talk begun with a focus on AWS Nitro which is segmented into modular components to rapidly launch new instances: Card, Security chips and Hypervisor.
VM’s can now function as bare metal servers with the flexibility and security of AWS cloud. Thedifferent characteristics of VM’s are: Processors, Memory, Local Disk Space and Network Performance. The format example would be: M5d.xlarge (family, generation, additional capacity, size) — d = local ssd disk.The different types of processors that exist and that are suitable for different use cases are: Intel, AMD, Aws (arm 64 bit) and GPU + FQGA accelerators.
There are also different types of instance types that exist:
- On demand (No contact, most expensive)
- Reserved Instances (Durational contract based VM’s to get a price reduction)
- Spot Instances (Not important if it goes down, if on-demand instance type demands go up) Up to 90% discount
New instance types are also to be noted: Z1d — High Calculation, C5n — Optimized for digital simulation and P3DN — Distributed training (Includes GPU nvidia tesla).
The recommendations given on how to choose a instance type where to:
- Benchmark and test different spot instance
- Read Systems Performance by Brendan Gregg
- Optimize the code or a faster processor may be needed
Conclusion
This is only a sample of the wealth of improvements and innovations that were presented at this AWS Summit. The speed of change (that « x4 » factor, contrast it with the usual « +x% ») is an evidence of the platform effect, that is enabled by a thoroughly designed architecture making heavy use of APIs, and its ecosystem of partners. From platform to winner-takes-all, there is a small step; fortunately competition is lively and the open source community thriving!
Co-written by : Régis Martin, Thomas Scott, Alfie Walker and Ahmed Bahet
