Data Center Links: August 24, 2017

Here are some (mostly) recent things I found interesting:

• Apple to build data center in Iowa.  The Des Moines Register reports that Apple will build a data center in Waukee, Iowa, a suburb of Des Moines. Following large data center builds from Microsoft and Facebook in the Des Moines area, and Google in Council Bluffs, Apple will fulfill project 'Morgan' in a newly annexed portion of Waukee.  My thoughts? Smart move by Apple (Iowa=awesome location), data center hub in the making (enormous build-outs by Microsoft, Facebook and Google in Iowa), and good for Waukee (there is already another totally awesome data center in town). #iowa  #iowabrag

• Druva nets $80 Million Funding round.  Cloud data and protection management company Druva announced $80 million of growth equity funding Tuesday, bringing their total raised to around $200 million. Funding was led by Riverwood Capital, with participation from Sequoia Capital India, Nexus Venture Partners, Tenaya Capital and most other existing venture investors. Druva said they will use the funds to "dramatically accelerate research and development, expand go-to-market efforts worldwide, and lead the industry in redefining how enterprises protect, manage, and use their data."

• Microsoft Acquires Cycle Computing. Microsoft announced it is acquiring Cycle Computing, a leader in HPC cloud computing orchestration. Microsoft says it will integrate the Big Computing capabilities from the CycleCloud product into Azure. 

• Microsoft launches Brainwave for real-time AI. At the 2017 Hot Chips Symposium in Cupertino Microsoft announced Project Brainwave, a new deep learning acceleration platform. Leveraging a large FPGA infrastructure inside of Microsoft, the new platform is built with three main layers: a high performance, distributed system architecture, a hardware DNN (deep neural network) engine synthesized onto FPGAs, and a compiler and runtime for low-friction deployment of trained models.

• Databricks secures $140 million. Unified Analytics Paltform company Databricks announced it has received $140 million in a Series D funding round, led by Andreessen Horowitz. Founded by the team who created Apache Spark, Databricks has raised $247 million to date, and will use the new funds to accelerate the investment in making artificial intelligence achievable for enterprise organizations with its Unified Analytics Platform. 

• Red Hat, Microsoft simplify containers for hybrid clouds. Red Hat and Microsoft announced an alliance expansion, with a new initiative aimed at enabling enterprises to more easily adopt containers. The companies say this includes "native support for Windows Server containers on Red Hat OpenShift, Red Hat OpenShift Dedicated on Microsoft Azure, and SQL Server on Red Hat Enterprise Linux and Red Hat OpenShift." 

• ORNL readies facility for 200 Petaflop Supercomputer. Oak Ridge National Laboratory is getting a data center ready for its Summit supercomputer, which is expected to deliver 200 petaflops - which is about twice as powerful as the world's current leader in computer performance. The data center includes a new 20 megawatt power and cooling plant and an expanded central energy plant for the campus. 

Data Center Links: May 11, 2017

Between company conference season and a lot of M&A activity recently, it is hard to keep up with the news.  Here are a few things I found interesting:

• Facebook breaks ground on 1M sq ft Altoona data center.  Because Iowa is the best place for data centers.... Facebook broke ground on its fourth building in Altoona - a 1 million square foot data center. With this building the Facebook data center campus will total 2.5 million square feet.

• Cisco acquires AI company MindMeld.  Just like many years ago when you had to keep a running list of who Cisco was buying, they are keeping a fast pace in 2017. Making it their third acquisition in two weeks Cisco dives deeper into AI by buying MindMeld, a San Francisco company that has developed a conversational platform based on natural language understanding (NLU). The NLU functionality will be integrated with Cisco's Spark platform. MindMeld, besides being on my own 'cool tech' list, has been recognized as one of the top companies leading the AI revolution.
• Could Cisco possibly acquire Nutanix, vArmour or SnapRoute next?

• Cray launches Supercomputers built for AI.  Cray has launched new CS-Storm supercomputers that are purpose-built for the most demanding AI workloads.  New CS-Storm 500GT and CS-Storm 500NX systems leverage NVIDIA Tesla GPU accelerators, which foster "up to 187 TOPS (tera operations per second) per node, 2,618 TOPS per rack for machine learning application performance."   I'll take two.  :)

• NVIDIA GPU conference.  AI, self-driving cars, healthcare, Volta GPU platform, deep learning software tools.....   Too many to mention; check it out here.

• Microsoft debuts Cosmos DB. At the Microsoft Build 2017 developer conference Microsoft introduced Azure Cosmos DB, a globally distributed database with five consistency choices. As a superset of DocumentDB Cosmos models include Strong, Bounded Staleness, Session, Consistent Prefix, and Eventual. 

• Apple announces $1B data center expansion in Reno. 9to5Mac reports that the Reno city council approved Apple's proposal for a $1 billion expansion to its existing data center. 

Data Center Links: May 3, 2017

Here are some (mostly) recent things I found interesting:

• Shaw explores selling ViaWest: Sources 'familiar with the matter' say that Canadian cable company Shaw Communications is exploring options for selling ViaWest. Shaw bought ViaWest three years ago for $1.2 billion.

• Amazon Strategy Teardown.  CB Insights has an amazing, detailed analysis of strategy at Amazon.  Company history, AI-as-a-Service, acquisition highlights, 'aggressive' growth plans, and initiatives by sector.  Wow.

• iRobot: Vacuuming up Microservices on AWS.  I like cloud architecture stories. This one just spoke to me -- about microservices and modern applications and services.  "Ben Kehoe, Cloud Robotics Research Scientist at iRobot, explains how they built a serverless solution to power microservices that scale to control millions of robots. You’ll learn how they automate and optimize AWS Lambda function and Amazon API Gateway deployments with AWS CloudFormation and Swagger, plus how they inject information during the deployment process to decouple architecture details from the code."

• Will Molten Silicon make Lithium-Ion 'Uneconomic'?  Australian company 1414 says it has  developed a molten silicon thermal energy storage system (TESS) that can store 500 kilowatt-hours of energy within a 70-centimeter cube. At 36x the capacity of a 14 kilowatt hour Tesla Powerwall 2 the company says it could build a 10 megawatt-hour plant for around AUD $700,000 (USD $528,000), or a tenth of the price of a Tesla battery-based project.

• MIT Mathemetician spins up 220,000 core Google Compute Cluster.  Google announced that MIT math professor and computational number theorist Andrew V. Sutherland crafted a 220,000-core workload on Google Compute Engine using preemptable virtual machine instances. Sutherland is already planning an even larger run of 400,000 cores.

• Path to Exascale. At a user forum U.S. Exascale Computing Project (ECP) director Paul Messina outlined the accelerated timetable with delivery of the first exascale machine now scheduled for 2021. The ECP is a collaborative effort of two U.S. Department of Energy organizations, the Office of Science (DOE-SC) and the National Nuclear Security Administration (NNSA).

Data Center Links, August 23, 2016

Data Center Links for 8/23/2016.

Here are some (mostly) recent things I found interesting:

• Avere and Cycle Computing integrate for hybrid HPC in the cloud.  Avere Systems and Cycle Computing announced a technology integration that enables hybrid high performance computing in popular public cloud computing environments. By combining the CycleCloud with Avere's vFXT Edge filer users can launch an Avere tiered file system on demand linked directly with the CycleCloud managed scalable compute nodes through cloud providers like AWS, Google and Microsoft Azure.

• U.S. DOE awards $34 million to protect power grid.  The U.S. Department of Energy has awarded $34 million to help protect critical infrastructures, specifically the smart grid. The funds cover 12 projects in the Office of Electricity Delivery and Energy Reliability's Cybersecurity of Energy of Delivery Systems (CEDS), and is intended to help develop new solutions to protect critical infrastructure in the energy industry.

• U.S. approves handover of IANA to ICANN. The National Telecommunications and Information Administration (NTIA) has approved a plan to hand control of the Internet Assigned Numbers Authority) contract to the Internet Corporation for Assigned Names and Numbers (ICANN). ICANN has run IANA functions - DNS, IP addresses and protocols - since incorporation in 1999. 

• NVIDIA Parker - SoC for autonomous vehicles. At the #hotchips conference NVIDIA announced Parker, a new mobile processor that the company hopes will power the next generation of autonomous vehicles. Built around the Pascal GPU architecture and NVIDIA's Denver CPU architecture, the company says it will deliver "up to 1.5 teraflops of performance for deep learning-based self-driving AI cockpit systems." NVIDIA CEO Jen-Hsun Huang delivered a supercomputer to OpenAI, a non-profit founded by Tesla's Elon Musk. 

• SKA Science Data Processor (big data meet big compute). A prototype part of the software system to manage data from the Square Kilometer Array (SKA) telescope has run on the Tiahnhe-2 supercomputer - currently the second best supercomputer in the world.   Deemed the world's largest science project, the SKA data processing system will ingest data from more than a quarter of a million antennas. 

• CyrusOne and the REIT sector. Seeking Alpha has a nice look at data center REIT CyrusOne and their record leasing quarter - and a bullish thesis on the company and its future. 

• Intel to Fab ARM.  I had to check that headline a few times - but yes, Intel Custom Foundry will now "offer access to ARM Artisan physical IP, including POP IP, based on the most advanced ARM cores and Cortex series processors. Intel also told of several foundry success stories from LG Electronics, Spreadtrum, Achronix Semiconductor, Netronome and Altera. 

SuperComputer 2008 Conference

SC08 (The International Conference for High Performance Computing, Networking, Storage and Analysis) has been going on this week and there are a number of interesting links as a result:

• Stacey Higginbotham at GigaOm points out that the conference is really (or has become) a Data Center conference.
• The Austin American-Statesman previews the conference and points out that Sun will bus people to their Ranger installation at UT, and Dell will bus people to UT's main campus to see Stallion, their advanced computer visualization system.
• Nortel 100G to carry critical network traffic for SC08 conference.
• MercuryNews.com reports that Intel, Cisco and Sun are helping develop a supercomputer for the Lawrence Livermore National Laboratory.
• NVIDIA announces Tesla "personal supercomputer"...... umm...... ok
• Oak Ridge National Laboratory selects DataDirect Network Storage to power one of the largest and fastest file systems in the world. "48 S2a9900 High Performance storage platforms deliver up to 240 Gigabytes per second of performance across 10 Petabytes". Wow!
• DataDirect Networks also won the HTCWire's Editors' Choice Award for best HPC Storage Product for 2008.

Finally -- I have a YouTube video to share. It's not SC08 related, but you would need a supercomputer to do some of the stuff this guy is talking about. It's the latest Google TechTalk, and even the title of the talk is over my head: The Liquid Fluoride Thorium Reactor: What Fusion Wanted to Be
Enjoy