Martin Beeby, leading developer advocate for Amazon Web Services (AWS), announced during a keynote presentation at AWS Summit in New York a new way to run workloads globally, AWS Cloud WAN, the “WAN” – which stands for Wide Area Network.
“It’s a single console to manage everything,” says Martin Beeby. Using Cloud WAN enables a business to run programs as a combination of AWS Availability Zones, including AWS “Local Zones” and AWS “Outposts”, deployments that are located at the inside companies’ own data centers, notes Martin Beeby, as well as many other AWS resources such as edge computing.
The one-day event was held at the Jacob Javitz Convention Center in Midtown Manhattan, New York. It was the second live US event, after the one in San Francisco. Martin Beeby replaced Amazon CTO Werner Vogels, who was scheduled to deliver the keynote address but was unable to do so due to weather issues. “Think of me as Werner Vogels’ Availability Zone,” joked Martin Beeby.
Configure all parts of a network, including private clouds
Cloud WAN is promoted as a way to easily set up and manage multiple parts of an AWS-based network around the world, including AWS’s “virtual private clouds,” or VPCs, as well as private cloud offerings on customer site. More details are available in a blog post by Sébastien Stormacq, a leading AWS developer advocate.
Customers using Local Zones include Finnish game maker SuperCell, author of Clash of clans, notes Martin Beeby. They use local zones in the United States to reduce the latency experienced by their American players. Financial exchange company Nasdaq uses AWS outposts to run AWS services in its own data center.
AWS also announced that Delta Airlines has chosen it as its “preferred cloud service provider”.
Focus on Graviton3 and Honeycomb
Martin Beeby spoke at length about Amazon’s line of custom computer chips, which includes “Trainium”, for machine learning forms of artificial intelligence; “Inferentia”, to speed up predictions with trained models and “Graviton”, designed to speed up general workloads.
The third generation of Graviton was unveiled at the end of May, “Graviton3”, and some AWS customers have transferred workloads to the new chip with favorable results in terms of cost reduction and performance improvement.
An example of using Graviton3 is Honeycomb dot io, which builds programs for observability. According to Liz Fong-Jones, Site Reliability Engineer and Developer Advocate at Honeycomb, the company was able to reduce costs by 60% compared to its previous use of AWS by running programs in Seventh Generation instances of AWS. , “C7G”, moving from fifth-generation instances running on the Graviton2 chip that Honeycomb previously used. “As many people as possible should adopt Graviton3 because it’s much more energy efficient, much more carbon efficient,” she told ZDNet after the keynote.
“Graviton3 has led to significant performance benefits,” continues Liz Fong-Jones. Honeycomb competes with many companies in observability and application performance monitoring, including start-up Lightstep. “Honeycomb is uniquely differentiated by speed and scale, and AWS enables us to achieve that speed and scale. »
“In terms of how we do it, it’s Graviton, Lambda and Spot Compute. These technologies have been fundamental to us,” she adds, referring to AWS serverless technologies, under the umbrella of AWS Lambda, and “Spot” instances, which consist of using the available EC2 capacity that Amazon offers prices up to 90% of normal AWS EC2 instances.
Describing speed and scale, Liz Fong-Jones tells ZDNet that “10 seconds [temps de requête] is the limit of what we consider acceptable, the median request time is less than 500 milliseconds, while our competitors can take 30 seconds, one minute, two minutes”.
While “competitors may say we return results immediately, those competitors pre-aggregate the data, which limits the dimensions you can query on,” says Liz Fong-Jones. In contrast, “Honeycomb uses a columnar index format, and we can give you any combination of fields on the fly.” This agile search is made possible by AWS Lambda. That’s important because observability really means asking “open-ended” questions, she added.
Reduce costs and increase energy efficiency of infrastructure
During the keynote, Martin Beeby urged developers to use AWS Lambda, both for cost reduction and infrastructure energy efficiency reasons. “You really should start looking at serverless. It can truly save you money and help us save the planet,” argues Martin Beeby.
Serverless technology is a form of compute-stateless microservice, which means it can handle individual requests for resources, such as individual database queries, without the overhead of an entire server instance. It thus saves computing power and costs for certain types of work.
Martin Beeby presented a serverless version of the Amazon Redshift data warehouse program. It is one of three parsers in a serverless version, the other two being EMR, the Apache Spark service, and a serverless offering of Apache Kafka, the event streaming service.
Martin Beeby’s talk was repeatedly interrupted by protesters shouting from the audience various denunciations of AWS, including “Stop hurting immigrants, stop contracting with ICE,” in reference to the Department’s Immigration and Customs Enforcement Division. of United States homeland security. The protesters were escorted out of the venue by Javitz Convention Center security.
Source: ZDNet.com