Last year I shared a blog post about Amazon Web Services’ (AWS) vision and key product announcements that I thought were important for new and existing AWS customers. AWS already provides a wide selection of services to customers with greatly varying needs. In light of this growing breadth and depth of customer needs, they continue to utilize customer feedback when improving existing services and creating new ones.
Similar to last year, instead of listing every update or new service (that would be a very long blog), I decided to highlight the vision and key concepts AWS shared throughout various keynotes and product announcements. For a complete list of all the updates and releases, click HERE.
Flexible Storage Options
As the diversity of data stored in AWS rises at a staggering rate, customers not only need performant storage solutions, but also a more feature-complete storage solution. Andy Jassy, CEO of Amazon Web Services, highlighted this concept in his keynote explaining that AWS’s vision is not only to provide cloud solutions but feature-rich solutions. Storage is one of the areas that has seen these enhancements. S3 Batch Processing is a feature that is in high demand. This feature would allow users to run batch operations with a single API request and save time for developers by not having to create their own custom solutions.
In addition to these new storage features, AWS also announced methods for cost optimization. By applying machine learning to object access patterns, AWS introduced a new tier of S3 storage called Intelligent Tiering, that allows AWS to choose the correct, most cost-effective tier based on how frequently that object is requested. For objects that are stored for retention, AWS released S3 Glacier Deep Archive which features even lower prices than currently available with Glacier.
One pain point I noticed while working with a variety of AWS customers was the inability to use all the security features that AWS provides, along with other third-party tools to establish a great system for monitoring security. This year, AWS Security Hub was announcement; this tackles the problem and provides a centralized view of compliance status and alerting.
In addition to security management, multi-account customers also struggle to implement standards across multiple accounts. I have helped clients build custom solutions to address this issue. However, last week AWS announced AWS Control Tower service that will automate a lot of the manual processes of creating secure multi-account AWS environments.
AWS also provides a variety of databases that customers can leverage to best meet their business needs. Last year, with the release Amazon Neptune, AWS was offering database options for relational, non-relational, and graph databases. To my surprise, AWS released new managed database services that are optimized for specific workloads. For IoT workloads or any other time series workloads, AWS released Amazon Timestream which allows for storage of a myriad of timestream events at a better cost than relational databases.
Additionally, Amazon released Amazon Quantum Ledger Database (QLDB). This database allows you to record a history of economic and financial activity or any other workloads that require maintaining an accurate history of the data. Andy Jassy mentioned that their vision is to provide freedom for application builders to use the tools that work best for their use case. Supplying these workload-optimized managed databases showcases AWS’s commitment to listen to customer requirements and provide the tools that are more efficient and cost-effective solutions.
Machine Learning Expansion
If you were to ask me five years ago whether AWS is a great provider for machine learning workloads, I would have responded with a caution stating that machine learning was not their strong suit. However, today, AWS has grown significantly in this area and provides a plethora of tools and services that accommodate machine learning workloads. First, AWS provides extremely large instances such as the P3 instance-family that allows for compute and GPU intensive workloads. While P3 instances are good options for running intensive machine learning workloads, they can also be sometimes cost prohibitive for some companies. This week, AWS announced Amazon Elastic Graphics feature that allows you to attach graphics acceleration to any instance size. This kind of instance configuration allows applications to be optimized for the exact amount of graphics acceleration needed for the workload while providing the cost benefits of not running extremely large GPU-based instances.
New managed machine learning services were also announced that allow AWS customers to innovate faster without having to reinvent the wheel. Amazon Textract is a service that uses enhanced character recognition to extract text and data from scanned documents and forms. In addition, Amazon Forecast makes predictions for time-series data. This service uses the same predictive models that Amazon retail uses to forecast their resources. Amazon also released their recommendation models as well in their newly announced Amazon Personalize service.
I can confidently say that AWS is now a great place to run machine learning workloads. From running your own models using AWS SageMaker or Amazon Elastic Graphics to leveraging AWS managed services such Textract, Polly, Translate, Forecast, Rekognition, or Amazon Personalize, AWS has a diverse set of machine learning capabilities.
With AWS moving so fast to meet customer needs, these were just a few key concepts identified from AWS re:Invent 2018. Each year I’m surprised to see how Amazon continues to adhere to changing customer requirements and providing diverse services.
If you have questions about any newly released AWS services (or anything AWS-related), let us know! Schedule a consultation with our AWS Experts. We’d love to chat with you about how 1Strategy can help your business with your journey into the AWS cloud.