AWS re:Invent - Compute, Network, and Data Center Design Principles
AWS re:Invent Andy Jassy Keynote – IT Application Services

AWS re:Invent Andy Jassy Keynote - Core IT Compute Features

FstestgrowthITAmazon Web Services (AWS) CEO, Andy Jassy re:Invent keynote was a mixture of his view on enterprise IT and many new services announcements. AWS is now a $13-billion-dollar business, growing at 55% year over year. Andy reported that AWS is now the fastest growing large enterprise IT company in the world. The next four fastest growing enterprise IT companies are using AWS as their primary cloud provider. AWS used this re:Invent conference to announce their readiness for supporting enterprise IT and software as a service workloads in addition to their success with supporting startup businesses. AWS is now services more than 1 million active non-Amazon customers and delivering 300 million hours of EC2 services monthly. AWS has commoditized most of the common IT infrastructure services as robust, easily accessible cloud services. In this post I will recap the the announcements of new and enhanced core enterprise IT compute features. In my next post I will recap how AWS is pivoting from low cost IT infrastructure provider to providing value added business IT services.

The first new capability announcements were a group of new EC2 instance types mostly taking advantage of Moore’s law. The enhanced instance type announcements were:

  • T2 –New instance types: T2.xlarge & T2.2xlarge – 2x memory – for more in memory processing workloads
  • R4 – memory intensive workloads – 2x everything
    • Storage Capacity Instance
    • Speed using DDR4 memory
    • L3 cache capacity
    • vCPU’s
  • I3 – replaces I2 instance type for I/O intensive workloads
    • 9x – 9x IOPS
    • 2x memory capacity
    • 3x storage capacity using NVM to replace SSD media
    • 2x vCPU’s
  • C5 – replaces c4 instance type for compute intensive workloads like AI, transcoding.
    • Featuring Skylake CPU’s
    • 2x CPU capacity
    • 2x CPU speed – move from Haswell to Skylake Intel CPU’s
    • 3x storage capacity moving to NVM from SSD media
    • 4x more memory

In addition, two new instances were announced introducing the availability of GPU’s to machine learning and artificial intelligence processing, and field-programmable gate arrays (FPGA) for offloading repetitive processing tasks.

  • New P1 Instance – Instance including everything need to leverage GPU. Full and share GPU (spot market) available
  • New F1 Instance type – Instance including everything needed to develop FPGA acceleration. FPGA Market Place ready to go at launch

The AWS EC2 catalogue is broad enough to meet the majority of enterprise workload types. Each of these instances can be launched in a few minutes. AWS has also recognized a class of use cases such as simple web, application, or basic Linux servers is needed. For these use cases an instance type with a minimum of features including basic storage, networking and operating system is needed. For this use case AWS announced the launch of their Lightsail service. With a few menus driven clicks you can now launch a functioning virtual private server (VPS). The current VPS service does not offer an easy upgrade path to any of the other EC2 instances like the Lightsail offering.

compute service portfolio is broad enough to service most enterprise IT workloads. One of the major challenges enterprise IT has always dealt with is optimizing, and reducing the cost of operating their existing applications. Most of applications created in the past decade run on VMware virtual servers and migrating those workloads to AWS has been disruptive and complex. AWS and VMware have announced a new partnership which will enable VMware workloads to run on dedicated bare metal EC2 instances by mid 2017. This will allow enterprise IT to take advantage of AWS scale and cost advantages while not having to modify their existing applications or operational procedures. 

During this part of his his keynote Andy Jassy is clearly making the case that enterprise IT should be fully embracing AWS as it’s primary cloud provider. Andy’s argument is many startups that threaten to disrupt established businesses today and the software as a service (SAAS) provider’s such as Workday and Salesforce that enterprises on today are already relying on AWS cloud services so you need too as well.  


Feed You can follow this conversation by subscribing to the comment feed for this post.

The comments to this entry are closed.