{"id":20641,"date":"2022-12-02T21:16:09","date_gmt":"2022-12-02T21:16:09","guid":{"rendered":"https:\/\/mailinvest.blog\/index.php\/2022\/12\/02\/analyst-highlights-zero-etl-and-processor-advances-among-key-news-from-aws-reinvent\/"},"modified":"2022-12-02T21:16:09","modified_gmt":"2022-12-02T21:16:09","slug":"analyst-highlights-zero-etl-and-processor-advances-among-key-news-from-aws-reinvent","status":"publish","type":"post","link":"https:\/\/mailinvest.blog\/index.php\/2022\/12\/02\/analyst-highlights-zero-etl-and-processor-advances-among-key-news-from-aws-reinvent\/","title":{"rendered":"Analyst highlights zero-ETL and processor advances among key news from AWS re:Invent"},"content":{"rendered":"

<\/a>\n
<\/p>\n

\n

One of the messages to emerge during the AWS re:Invent gathering<\/a> in Las Vegas this week is that Amazon is declaring war on the need to extract, transfer and load data.<\/p>\n

The ETL process has been a sore point<\/a> within information technology organizations because of the time and expense necessary to extract and move information from data sources not optimized for analytics. AWS addressed the ETL issue with news<\/a> this week that Amazon Aurora would now support zero-ETL integration with Amazon Redshift, enabling near real-time analytics and machine learning on petabytes of transactional data.<\/p>\n

\u201cThere was the zero-ETL that they talked about, and it\u2019s a little bit of a game-changing moment,\u201d said Andy Thurai<\/a> (pictured), vice president and principal analyst at Constellation Research. \u201cYou don\u2019t have to move the data around, use the data where it is.\u201d<\/p>\n

Thurai spoke with theCUBE industry analysts John Furrier<\/a> and Dave Vellante<\/a> at AWS re:Invent<\/a>, during an exclusive broadcast on theCUBE, SiliconANGLE Media\u2019s livestreaming studio. They discussed major themes and messages from keynote presentations this week. (* Disclosure below.)<\/em><\/p>\n

Instances for every workload<\/h3>\n

Another theme from the re:Invent conference involved Amazon\u2019s ongoing commitment to processor development. The company made a number of updates<\/a> to its custom chip lines this week for analytics scaling and data management.<\/p>\n

\u201cWhat they announced, which is somewhat interesting, is they have 700 different instances geared toward every single workload,\u201d Thurai said. \u201cSome of them specifically run on Amazon\u2019s Inferentia2 and Trainium Trn1 chips. The numbers have to be proven, but they claim it could be anywhere from 40% to 60% faster.\u201d<\/p>\n

The net effect from the lengthy list of announcements<\/a> made by AWS this week was to reinforce a view that the cloud giant intends to provide the broadest array of tools for enterprise computing needs.<\/p>\n

\u201cThey want to become the platform that, when you come in, it\u2019s the only platform you would use from model training to deployment to inference to\u00a0 versioning to management, the whole nine yards,\u201d Thurai said. \u201cThat\u2019s the angle they are trying to take; it\u2019s a one source platform.\u201d<\/p>\n

Here\u2019s the complete video interview, part of SiliconANGLE\u2019s and theCUBE\u2019s coverage of AWS re:Invent<\/a>:<\/p>\n