AWS continued its methodical push this year to gobble up more of the IT landscape, with a mix of forward-looking tools and upgrades to link and expand its existing portfolio of services.
Microsoft and Google did their part to catch up with AWS in 2017, but both cloud providers worked against a moving target. AWS updates pushed into new IT segments as the cloud provider took steps to fill gaps in its strategy around hybrid cloud, customers operating at a global scale, machine learning and containers.
It was another year of gradual improvements rather than the massive advances that embodied the earlier years of AWS, but there was still plenty for enterprise users to digest.
Among the most popular advancements in 2017 was the decision to finally cede to the popularity of Kubernetes with Elastic Container Service for Kubernetes.
Logicworks, an AWS managed services provider in New York, built its own tooling to support Kubernetes for its clients, and is more than happy to relinquish those duties to AWS going forward.
“Kubernetes was not easy to do on AWS,” said Jason McKay, CTO and senior vice president at Logicworks. “A lot of the open source tooling was not built to that network topology, and that was a real challenge.”
AWS updates reach into serverless, databases
Beyond its nod to containers, Amazon put even more attention on serverless frameworks such as Lambda and the growing list of services with which it interacts. It remains to be seen if either model will eventually dominate the cloud landscape and supplant VMs, but many observers expect companies will choose the serverless path to optimize their workloads after initial lift-and-shift migrations.
In fact, serverless frameworks have improved so much that providers such as Logicworks make it their default option internally and are seeing more traction with their customers. With AWS Step Functions and other upgrades, these services fit together better to provide true elasticity while they keep the focus strictly on the apps, McKay said.
Jason McKayCTO and senior vice president, Logicworks
“It’s really a continuation of what’s been happening. AWS is no longer an infrastructure-as-a-service cloud,” he said. “It’s a platform in which you develop your application and with serverless you’re inserting code into the platform without having to manage the infrastructure.”
Databases got plenty of attention this year as AWS executives continue to beckon corporations away from legacy systems, and skewer Oracle, too. Aurora and DynamoDB added capabilities to improve speed, consistency and uptime on a global scale. Multi-Master, an AWS feature currently in preview for both services, generated lots of interest because it can create multiple read/write masters in different availability zones.
New instance types addressed a range of uses, including compute-optimized machines for high-performance computing and storage-intensive VMs for big data applications. AWS continued to expand its data center footprint with several new regions, including one in France, second locales in China and GovCloud, and plans to expand to the Middle East in 2018.
Outages and exposed data are 2017 lowlights
Of course, it wasn’t all smooth sailing for AWS this past year. The major Simple Storage Service (S3) outage caused by human error took down scores of workloads in the U.S. East-1 region in February. The downtime cost businesses an estimated $150 million and hit particularly hard for companies that relied heavily on that region and didn’t have appropriate failover options in place.
However, the S3 outage wasn’t as detrimental as some broader outages that were more common in the past. It’s telling that 10 months later the incident barely registered with those interviewed for this article.
“People were upset, but I never heard one case where someone was like, ‘I’m out,'” said Adam Book, principal cloud engineer at Relus Technologies, an AWS consulting partner in Peachtree Corners, Ga. “Some wanted some redundancy and even some enterprises talked about redundant cloud solutions, but they all hit their SLA.”
Security was also a persistent problem — not so much for AWS itself, but for the steady stream of customers who left data publicly exposed. AWS has historically been hands-off with its customers, but it’s clear that the din of embarrassing incidents got the company’s attention. AWS reminded customers to secure their buckets and put default encryption policies in place, as well as alerts to notify users when their data is publicly exposed.
Other AWS security updates included more tools to help users secure their applications on top of its infrastructure. Among them were Amazon Macie, which brings machine learning to recognize personally identifiable information and alerting users about abnormal behavior involving that data, and GuardDuty, a fully managed service for threat detection.
“Amazon decided in 2017 it needed to own the governance process for their platform,” said Erik Peterson, a longtime AWS user and co-founder and CEO of CloudZero, a Boston startup that specializes in cloud security and DevOps.
Features extend beyond AWS cloud
AWS continued to focus on enterprise demands, particularly around hybrid cloud. The much ballyhooed VMware on AWS service became generally available in August, for corporations to move a full VMware stack onto AWS infrastructure. It’s unclear how seamlessly those VMware workloads will link to AWS tools as the offering matures, but in its early days the vast majority of the application has centered on disaster recovery.
Other initiatives pushed AWS farther afield from its own data centers. A partnership with Red Hat brings management of cloud resources in-house via OpenShift, while other efforts seek to put Lambda on edge devices and use Application Load Balancer to route traffic to private data centers.
Machine learning received plenty of focus too, particularly since it’s an area where AWS has lagged. SageMaker was the most intriguing of those initiatives, as the service aims to make machine learning approachable to more types of developers.
Some AWS updates connected services across regions as well, a shift for a company that has historically tried to wall regions off from each other for security and uptime purposes. New capabilities in AWS CloudFormation and CloudWatch make it easier to manage changes across regions and accounts, more End Points were added to Amazon Virtual Private Clouds (VPCs) and AWS Direct Connect can now link VPCs across regions.
Many of these AWS updates won’t generate a lot of headlines or get top billing at a conference keynote, but they’re plumbing upgrades that make it easier to work with the platform for day to day operations, Book said.
Not one to rest on its laurels, AWS also continued to add features that push the cloud platform further into ever-expanding subsets of enterprise IT. Alexa for Business incorporates the voice assistant into corporate workspaces. Amazon Connect is a contact center service for customer support centers. AWS Single Sign-On puts Amazon at the center of management of accounts on AWS and third-party applications. And Amazon Chime adds unified communications service to the AWS umbrella.
But with the steady stream of AWS updates and new services, it’s also getting harder to navigate which services to use, especially for new customers.
“[AWS needs] to really take a step back and ask themselves: Are they presenting all their services in the best way possible?” Peterson said. “I talk to people who just started on Amazon a bunch and they’re completely overwhelmed.”
Trevor Jones is a senior news writer with SearchCloudComputing and SearchAWS. Contact him at firstname.lastname@example.org.