Amazon Web Services (AWS) flexed its market-leader muscle in Las Vegas last week, announcing new cloud services and elbowing into on-premises datacenters. The announcement laid bare the challenge to both customers and competitors that models must change in order to keep pace.
The 2018 iteration of its re:Invent user conference saw the subsidiary of the US retailing-and-logistics behemoth make a concerted effort to lure customers to its platform and to distance itself from rivals.
In the cloud, where AWS owns half the global marketOpens a new window of as-a-service infrastructure, expanded compute, storage and analytics services are aimed at helping customers make more of emerging technologies like artificial intelligence and machine learning – all of it with enhanced security.
In the datacenter, the rollout of custom-designed hardware and cloud software updates helps those keen to maintain control of systems in-house also take advantage of that same compute functionality, with faster rates of data transfer, as they adopt hybrid cloud structures.
Platform Adoption Endeavors
AWS is simultaneously giving serverless computing a boost with open-source management tools designed to help developers combine the flexibility of container-based applications with virtual machines that increase back-end efficiency.
By expanding capabilities that span cloud and datacenter, AWS is making it easier for enterprise users to incorporate cloud as a key part of their operations. The Elastic Compute Cloud (E2C) platform that permits users to “rent†functionality is seeing numbers and types of instances grow. As a result, rivals will need to offer ever more customized processing, storage, memory and networking combinations to fit customer use cases.
AI is a pivotal player in such applications, as evidenced by the SageMaker tech AWS debuted at last year’s re:InventOpens a new window that featured a host of new applications like ML training with limited data, a compiler that optimizes software libraries and hardware architectures for faster processing, an automated data-labeling function and a marketplace repository for ML algorithms.
AWS said it will deliver E2C capability in customer datacenters beginning in 2019 with a service called OutpostsOpens a new window . The rollout will see the company install and maintain servers, and the software to run them on-premises, in a bid to lift more work to its subscription platforms. Outpost offers AWS an integration advantage against vendors Dell EMC, HPE and cloud rival IBM in hardware. It also mirrors existing software from Microsoft (Azure Stack) and Oracle (Cloud at Customer).
A Ripe Selection of Targets
The enhanced database capabilities announced at re:Invent also are a challenge to Oracle, whose software and services have long underpinned the AWS public cloud. In addition to expanding its Redshift data warehouse, AWS officials said the company transition of workloads to its Aurore and DynamoDB offerings should be completed by mid 2019.
Changes to Aurora allowed customers to replicate the MySQL database across AWS regions. Meanwhile, AWS is upgrading the Dynamo DB for documents that can handle up to 20 million requests per second with an automated read/write management capability, and gracing users with a pay per-request for the service.
Others include Timestream, for collecting and accessing time-sequence data, and a QLDB for data generated and contained in distributed ledgers. In on-premises systems, newly-released file servers for Windows and open-source Lustre are designed to reduce latency with solid-state drives.
The open-source effort extends to Firecracker, a management system for virtual machines in serverless computing environments. Built on the technology that lets developers forego infrastructure considerations with creating and deploying applications in the AWS cloud, Firecracker extends that capability to multi-tenant container environments.
AWS is taking aim at storage providers, as well, with its S3 Glacier Deep Archive using silicon to match tape in rates of degradation for infrequently accessed data, but at a fraction of the cost. Other additions to the Simple Storage Service (the S3) include automated tiering based on classes and access to further cut costs, batch processing for object data, and automated syncing of data held in the cloud and on-prem.
That bridge also is addressed in security offerings that let users better manage and view data across platforms and compliance regimes. Automated aggregation via the Security Hub lessens complexity, while a Control Tower helps in provisioning work among teams of developers.
Looking forward, some of those teams appear likely to be do so both on the AWS platform and its chip architecture. Officials said they hope to begin producing specialist silicon for ML applications in the coming year, news that demonstrates the breadth of the company’s cloud ambitions.