Sign up to save your podcasts
Or
Share StorageReview.com
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
StorageReview.com
By StorageReview.comTechnology
StorageReview.com episodes
- Podcast #111: HPE GreenLake Deep Dive
This week Brian takes us on an HPE GreenLake deep dive with Omer Asad. Omer is SVP & GM of the $1.4 Billion Data Storage, Data Services SaaS, and HyperConverged business at HPE Storage and is responsible for Engineering, Product Management, and Operations for all Data and SaaS services in Storage and Data Infrastructure platforms.
This week Brian takes us on an HPE GreenLake deep dive with Omer Asad. Omer is SVP & GM of the $1.4 Billion Data Storage, Data Services SaaS, and HyperConverged business at HPE Storage and is responsible for Engineering, Product Management, and Operations for all Data and SaaS services in Storage and Data Infrastructure platforms.
A key component of GreenLake is the Alletra product portfolio. HPE Alletra delivers agility and simplicity for every application from edge to cloud and can be provided as a service, or a fully managed turnkey offering, all with the right mix of subscription and usage-based services. Recently, we ran a piece on the Alletra 5000, the latest in the Alletra portfolio.
Prior to becoming a member of the HPE Storage team, Omer was SVP & GM at Nimble Storage, acquired by HPE in 2017. Omer has also led teams at Pure Storage and Riverbed Technology. To say he has a deep technical understanding of customer challenges would be an understatement. In addition to storage, Omer is well versed in cloud and edge solutions.
More on HPE GreenLake from the June launch.
00:00 Introduction
- Recap around HPE Discover
- GreenLake service introduction
- HPE as-a-Service meaning
- Differentiation
04:40 What is GreenLake?
- Cloud consumption and operation model
- What does a typical customer look like?
- Move to Hyperscalers
10:00 Advantages and simplicity
- Public cloud vs. as-a-Service
- Rate card attributes
- Keep the rates simple
17:00 GreenLake overprovisioning for growth
- Determining how much hardware to deliver
- Importance of Infosight
- Sizing is performed through Infosight data simulation
25:00 Physical vs. the planning
- The timeframe from planning and worksheets to delivery
- Determined by the subscriptions
- How the hardware gets delivered
- Turning up the service
30:00 Provisioned before it gets there
- Fleet service
- Lifecycle environment
- Maintenance
- HPE maintains a state machine for every customer environment
- IT can do important work when subscribing to GreenLake
36:00 Customer owns the consoles
- Allowing HPE support to maintain and support
- What is the size of GreenLake
- Built for all customers
- Affordability – Rate Cards are available on partner sites
40:00 How does GreenLake fit the retail market
- Home Depot is a reference model
- Retail is a perfect fit
- Aruba is a crucial part of edge locations
- A rack can be built for most environments
45:00 Unexpected benefits of GreenLake
- Simplicity for operations and procurement
- Innovations for GreenLake
- Application frameworks
51:00 Wrap Technology – cloud and edge
Podcast Video
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #111: HPE GreenLake Deep Dive appeared first on StorageReview.com.
53 min - Podcast #110: The Rising Importance of Storage Accelerator Cards
For this podcast, Brian connected with Pliops Global VP of Products and Marketing, Tony Afshary. Brian has known Tony for a long time, so this is an effortless conversation with great insight into Pliops, accelerator card technologies, what is driving the market, and what the future might hold.
For this podcast, Brian connected with Pliops Global VP of Products and Marketing, Tony Afshary. Brian has known Tony for a long time, so this is an effortless conversation with great insight into Pliops, accelerator card technologies, what is driving the market, and what the future might hold.
Tony has been in the storage and accelerator technology sector for many years, working for companies like Intel, LSI, and Seagate before joining Pliops.
StorageReview did a detailed, in-depth review on the Pliops XDP recently and found it easy to install and use. The performance numbers were solid. Even giving Software RAID all the benefits available, the Pliops XDP far exceeded expectations based on real-world results.
There is a lot of information in this podcast, focusing on a hot product set. Certainly worth a listen.
00.00 Introduction
- What the next generation of accelerator cards can do for computing and challenges
- What problems are Pliops and their competitors trying to solve
- The gap between what the CPU and Software can do
- Bottlenecks when accessing NVMe drives
- The need to offload tasks from CPU to accelerators
05.00 Evolution from RAID cards to Accelerators
- RAID card design
- Storage drive technology created the need for new architecture
- Discussion around Pliops hardware – visual
- How the Pliops hardware works
- Design
- Key Value APIs
11.00 Data Flow
- How blocks flow through the card
- What happens before data is written to SSD
- Why Pliops is so effective and efficient
15.00 Dense NAND technology
- Pliops design
- works with any vendor
- the bigger the drive, the better
- the cheaper the drive, the better
- customer hesitation in going to larger drives
- Things to avoid with NAND
20:00 SR Tests with Pliops and Solidigm drives
- Simple to install
- Comparison between RAID5 and RAID0
- Results
- Why the need for a new ASIC?
- Cost, density, and geometry are positive with ASIC
- Flexibility
- What’s with the Lane?
25.00 More about Lanes
- Lanes make a difference in performance
- Managing fan-outs on a system
- Form factor impact
- CXL 2 v 3
30.00 AMD v Intel
- Both are okay with Pliops
- Interest in ARM technology
- No data on when applications will run on ARM
35.00 VMware support
- Orchestration and containers are the requirements
- Virtualized environments need support
- How do prospects check out Pliops
- access to Pliops lab
- most are deploying drives in their environment
- partnering with phoenixnap
40.00 Wrap up
Podcast video on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #110: The Rising Importance of Storage Accelerator Cards appeared first on StorageReview.com.
42 min - Podcast #109: Direct-on-Chip Evaporative Liquid Cooling
There continues to be increasing interest in liquid cooling technology but also trepidation in the unlikely event of some sort of leak or spill. Enter ZutaCore HyperCool2 direct-on-chip Enhanced Nucleation Evaporator (ENE), a single, closed-loop two-phase, waterless liquid cooling solution that yields unparalleled heat dissipation at the chip level.
There continues to be increasing interest in liquid cooling technology but also trepidation in the unlikely event of some sort of leak or spill. Enter ZutaCore HyperCool2 direct-on-chip Enhanced Nucleation Evaporator (ENE), a single, closed-loop two-phase, waterless liquid cooling solution that yields unparalleled heat dissipation at the chip level.
Brian invited Udi Paret, ZutaCore’s President, to discuss the technology around this new waterless liquid cooling system and how the data center mindset is changing with more movement to alternative cooling technologies. The benefits of liquid cooling systems are hard to ignore.
The ZutaCore HyperCool2 technology is revolutionizing the market by alleviating cooling boundaries at the chip level, server, rack, POD, and data center levels. Unlike water-based solutions that carry the risk of IT meltdown, HyperCool2 leverages a safe, non-conductive, refrigerant. It is a complete hardware system, enhanced by a software-defined-cooling platform resulting in a low-pressure system that triples computing densities on a fraction of the footprint.
Udi considers himself a strategic and operating executive with a demonstrated international history of success, leading Fortune 500 divisions and Venture Capital start-ups across a wide spectrum of industries like IP networking, data storage, and data center management in the enterprise, System Integrator, and renewable energy domains.
Prior to taking the helm with ZutaCore in 2019, Udi was engaged with the company as an advisor from 20017 through 2019. Udi continues to perform duties as an Adviser with Next Energy Technologies, a role he has held since 2014. Next has developed proprietary organic semiconducting materials — Soluble Small Molecule Organic Photovoltaic (SSM-OPV) — that are earth-abundant, low cost, and non-toxic.
Udi brings a lot of data center and cooling technology knowledge to this podcast. He also recognizes the need for companies to get up to speed on energy efficiencies and power management in small to massive data centers. This is a technology that will gain traction, so give this a listen.
Video short of the ZutaCore solution
ZutaCore site
Video pod on YouTube
Show notes
00:00 Introduction
- A quick intro to cooling liquid v air
- Perfect storm
- Accelerators driving thermal challenges
- Reaction from Dell Technologies World
- Brian runs through the solutions available
05:00 Where ZutaCore fits
- Overview of implementation
- Much like single-phase
- The difference is the liquid
- Visual of plate
- Delivery mechanism is fundamentally different
10:00 Comparisons
- Walk through the functions of the vapor and liquid pipes
- Available solutions from small in a rack, to end-of-row
- Using gravity to move liquid
15:00 Keeping it cool
- How ZutaCore keeps the solution so small and compact
- Energy efficient
- Steps to swap normal heat sink with ZutaCore
- Going beyond CPUs
- What is the go to market for ZutaCore
- Developing ecosystem including large SIs
20:00 Talking about the loop
- Scary concerns
- What happens if there is a leak?
- System designed to be leakage-free
- Should a tube get cut, the system would be shut down for that tube
- ZutaCore eliminates the need to add alternative safety mechanisms
- Easy to install with flexible tubing
25:00 System maintenance
- Not typical for liquid to need topped up
- Because it is closed loop and the make up of the liquid keeps it in tact
- Disposing of the heat
- Choices are determined by environment
- Heat injection units
30:00 Data Center envelopes
- ZutaCore can be integrated in existing air cooled data centers
- Power efficiencies are a differentiator
- Wide scope of solutions for different data center environments
- Technology is driving alternative cooling options
- Adoption for greener initiatives
35:00 AI is driving hardware growth
- To keep up with AI and ML server farms are growing
- That means more power to push air or retrofit a liquid solution to cool the hardware
- Providers must be able to deliver and service cooling solutions
- ZutaCore serves the HPC, edge and SMB markets
- All environments have a need for cooling
- Edge deployments are prime environments for ZutaCore
40:00 Mainstream vs. Hyper Scale markets
- Brian talks about the SR lab and the challenges around cooling
- ZutaCore’s simplicity makes it attractive to different groups
- Eliminates forklift upgrades
- No changes to processes
45:00 Economics
- There is an investment
- More efficiences
- Rationalize financials
- CapEx and OpEx savings for near term and long term projects
- Software Defined Cooling based on data collected within the data center
50:00 Gathering data to make systems more energy efficient
- ZutaCore sits on many endpoints to collect data
- Helps to lower power and make systems more efficient
- Control system to guarantee 70 degree of heat coming out
- Huge benefit from both hardware and software
- Collecting existing data will make turn around simple
55:00 Wrap up
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #109: Direct-on-Chip Evaporative Liquid Cooling appeared first on StorageReview.com.
58 min - Podcast #108: HCI at the Edge
Brian invites Jeff Ready to talk about //Scale Computing and their HCI product and edge computing during this very informative podcast. StorageReview.com recently profiled //Scale Computing and created a short youtube video showing the steps to set up the //Scale Computing HCI cluster. Let’s say it works as advertised.
Brian invites Jeff Ready to talk about //Scale Computing and their HCI product and edge computing during this very informative podcast. StorageReview.com recently profiled //Scale Computing and created a short youtube video showing the steps to set up the //Scale Computing HCI cluster. Let’s say it works as advertised.
Jeff describes himself as “a high-tech executive with a passion for entrepreneurship.” He is the founder and CEO of //Scale Computing, delivering edge computing, virtualization, and hyperconverged solutions. Jeff continues to lead after 15 years through multiple rounds of funding, explosive growth, and many industry changes.
Jeff is passionate about entrepreneurship and startups in technology and non-tech ventures. To keep himself busy, he sits on the board of three companies and is chairman of the GEO Foundation, a non-profit organization that creates innovative schools across the country in the communities that will benefit the most. And he is co-founder of Centerpoint Brewing Company in Indianapolis.
After watching this podcast, it will be evident that Jeff is passionate about //Scale Computing and technology. This is an interesting conversation, and Jeff is forthcoming on all the topics. The podcast will seem to fly by, but if you need to hop around, the timestamp is listed below.
00:00 Introduction
- Centerpoint Brewing Company
- Jeff is Co-founder
- Tech vs. Brewing
- Who is //Scale Computing
- Simplifying IT infrastructure
- Jeff’s vision for the company
- Create a system that can heal itself
- Build a system as reliable as the AS400
- Create a company as successful as IBM
10:00 Categorize //Scale
- How did Jeff communicate what //Scale did
- The initial product was storage
- Integrating open-source Kernel-based Virtual Machine (KVM) into the storage system
- Why not run the applications on that system? And Hyperconvergence is born!
- How //Scale coined the phrase Hyper-Converged Infrastructure
- Differentiating from other vendors
16:00 Disaggregation
- Open source community
- Getting into open source projects
- Shifts in workloads
- Open source projects are typically difficult to manage
20:00 It should be easy to use
- //Scale may not be the first to develop products because it needs to be wrapped into an easy-to-use system
- The system needs to take care of itself
- Everything is NOT going to the cloud – now we know that
- One size fits all rarely happens
25:00 Mid-Market focus
- Sometimes the mid-market lags behind
- Blue collar mentality in mid-market businesses
- Making the most of available tools
- What pieces go to the cloud and what stays on-premises
- Point use cases that make sense to move to the cloud
30:00 The abstraction
- Use case example
- Metalworks
- LAN traffic
- Industrial edge
- How open source helps mid markets
- Affordability
- How did //Scale weather the storm?
- //Scale was always different
- The leading competitor was VMware
- Keeping it simple
39:00 //Scale messaging
- It was complex selling into a VMware market
- //Scale was successful because of the simplicity
- Looking for Operating System alternatives
- With the focus on edge, //Scale was the right solution
44:00 Broadcom purchase of VMware
- Companies will think about it more before signing a new contract
- The purchase was not for R&D
- How will VMware support the smaller customers
- Using IBM as an example
- MS came in with client/server and took the business
- Customers might be nervous about renewing contracts
- Support is vital to //Scale
- Spending more on support than other companies
- A real person answers the support line
52:00 //Scale has a different approach
- Apple would be jealous of //Scale’s customer base
- //Scale is still an unknown to many
- Customer fear of acquisition
- //Scale has survived while larger competitors were bought up
- Taking an active position in supporting the customer community
1:00:00 How are customers using //Scale
- Existing users will respond immediately when a question pops up in the community.
- There is a sense of community
- It works as advertised
- The new edge use case is opening the door for //Scale
1:06:00 Wrap-up
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #108: HCI at the Edge appeared first on StorageReview.com.
1 hr 7 min - Podcast #107: Q and A From The Lab With Kevin OBrien
Brian invited Genius Lab Guy Kevin O’Brien to step out of his comfort zone and answer burning questions from the community and a few topics pulled from social media.
Brian invited Genius Lab Guy Kevin O’Brien to step out of his comfort zone and answer burning questions from the community and a few topics pulled from social media.
Kevin has been with StorageReview.com for over 11 years, with his first role as a Managing Editor, directing the Enterprise Test Lab and finding ways to test the next generation of storage devices in new and challenging ways. He would step out of the lab and put on his managing editor hat to review content for our site’s enterprise and consumer sections. After a couple of years in the managing editor role, Kevin moved to Lab director overseeing the day-to-day operations of the StorageReview.com Test Lab, building company relationships to grow lab resources, and researching new testing methodology to incorporate into reviews.
In addition to running the lab, performing evaluations, developing test scenarios, and working with vendors, Kevin takes time to review some of the content for technical accuracy before it gets published.
Kevin rarely goes one-on-one with Brian, so give this one a view.
00:00 Introduction
- Recap of vendors that have come through the lab
02:00 Question #1
- What is the best NAS?
- Synology vs QNAP vs TrueNAS
- A matter of trust
- What works well with modern devices
- Security gotchas
- Features
- Innovation
11:00 Question #2
- Caddies
- Why do we need them
- Cost
- Smart designs
14:00 Question #3
- Advice on Networking
- What’s with the fiber?
- Cables
- The environment will dictate cable selection
17:00 Question #4
- Lenovo P620
- Why not build your own?
- Fans on DRAM
- Disadvantages to Build Your Own
- System operating models
20:00 Question #5
- Dell ME5
- Buy or build your own?
- Costs
- NVMe is ready for primetime
- Storage at branch offices
- The lower end of SMB
- Pricing from big players vs. smaller vendors
- HPE GreenLake
- On-Premises vs. Cloud
- If your costs are inching toward $10k-$15K, it might be time to look at enterprise kit
28:00 Question #6
- Why are there so many batteries in the office?
- And why are there so many portable power station reviews on the site?
- There are more systems
- More interest from the community
- Power grids are not as reliable
- More reliance on portable energy
- More use cases
- Edge
- Retail
33:00 Wrap-up
Let us know if you like this format, and we will do more in the future.
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #107: Q and A From The Lab With Kevin OBrien appeared first on StorageReview.com.
34 min - #106: AWS Snow Family Deep Dive
Brian calls on Wayne Duso, Vice President, AWS Storage, Edge, and Data Governance, to join him in this podcast and talk about data transfer and the power of the AWS “Snow” offerings. You might be amazed at the computational power, not just data transfer capabilities, that’s available within the Snow family.
Brian calls on Wayne Duso, Vice President, AWS Storage, Edge, and Data Governance, to join him in this podcast and talk about data transfer and the power of the AWS “Snow” offerings. You might be amazed at the computational power, not just data transfer capabilities, that’s available within the Snow family.
Wayne describes himself as a curious and articulate entrepreneur, investor, pioneer, builder, and leader who’s passionate about sponsoring rising-leaders, building teams, launching products, and growing businesses that deliver highly disruptive value for customers and benefit the communities I have the privilege of serving.
As a long-time Amazon employee (he’s been there almost ten years) and AWS contributor, Wayne has a deep understanding of AWS products, especially where it concerns data movement and storage.
Snowcone on the ISS
AWS and Axiom Space are collaborating on developing a more efficient way to analyze data from Axiom Mission 1 (Ax-1) via AWS Snowcone SSD. AWS Snowcone is a small, rugged, and secure device offering edge computing, data storage, and data transfer on the go in an austere environment with little or no connectivity. Ax-1 is the first of several private space missions to ISS planned by Axiom Space, with the ultimate goal of building the first commercial space station.
We ran an article in June that described the collaboration between Amazon and Axiom Space Remotely Operate AWS Snowcone on the International Space Station (ISS). We also have a video of the Snowcone setup on our Instagram page.
More information on AWS Snow products can be found on AWS Snow Family.
This is a very informative discussion with plenty of detail about AWS’s Snow products, Edge Data, and Data Transfer options. If you want to skip around, the timestamps are listed below.
Timestamp
00:00 Introduction
- Discussion around Snow product
- Edge computing
- What problem was being addressed
- The journey to move massive amounts of data
- Created Snowball
05:00 How it works
- Customers wanted to process the data before transfer
- Service evolved to collect and process data
- Creation of Edge device
- Data migration expectations
10:00 Outposts and Snow
- Outpost systems are intended to extend an AWS region
- Snow is further out on the edge but connected to the Internet
- Snow devices at top-secret sites
- How that works with AWS and ingesting data
- Economics
15:00 Transforming business to Edge Compute
- Payment structure
- Supply and demand
- All about logistics
- Inventory management
20:00 Filtering services
- What is the process
- Use signals from customers to aid in planning
- Planning is done every year, and it changes every year
- What applications are being built
- How those applications are built
- And use cases
- Adding IoT and RDS to devices
- Flexibility in design
25:00 Hardware design principles
- Hardware shared between service groups
- Limit the number of part numbers
- AWS is a holistic environment
- Applications can be tested on a virtual machine
- Can be run in-region
30:00 Management
- Monitoring
- Reporting
- Edge Console
- Accessible via API
- Snowmobile – aptly named?
- Logistics for Snowmobile, Snowball, Snowcone
35:00 How to connect Snowmobile to the datacenter
- Simple explanation
- Make it simple so there is no need for professional services
- Lots of planning put in place
- Axiom application in ISS
- Moving photographs
40:00 Running Snowcone on ISS
- What did the ISS get
- Nothing special
- Same Snowcone as displayed on AWS with special tape wrapping
- How the data is protected
45:00 Security
- RAID protection
- AWS Datacenter
- ISS Snowcone is still in space
46:00 Wrap up
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post #106: AWS Snow Family Deep Dive appeared first on StorageReview.com.
49 min - Podcast #105: Infrastructure Trends from Edge to Core
Storagereview.com has always had a strong relationship with Supermicro with engagements in hardware testing, integration, and multi-vendor environments. There seems to be a Supermicro server somewhere in the lab. Supermicro has been at the forefront of hardware and software development innovation.
Storagereview.com has always had a strong relationship with Supermicro with engagements in hardware testing, integration, and multi-vendor environments. There seems to be a Supermicro server somewhere in the lab. Supermicro has been at the forefront of hardware and software development innovation.
This week, Brian invited Supermicro’s Michael McNerney to discuss the company’s vision for the future and a look in the rearview mirror. Michael is Vice President of Marketing and Network Security with Supermicro, so he is well-positioned to discuss what Supermicro’s vision is for the future and how technology drives the development of data center servers and edge markets.
The discussion on this podcast covers hardware and software, how it affects server design, private-cloud versus public-cloud, and what drives innovation. Michael brings his deep knowledge of the server environment and discusses Supermicro’s continued innovation path.
The timestamp for each section is found below if you want to hop around.
00:00 Introduction
- Michael’s focus at Supermicro
- Security is a primary focus
- Supermicro is prioritizing internal and external security
- Attack vectors
05:00 Modernizing
- Accelerating innovation
- What is exciting to Supermicro
- Hardware development is moving quickly
- Trying to keep up with the enterprise requirements
- Trying to arbitrate for our customers
- Keeping up with innovations
10:00 New form factors
- Considerations for new hardware
- Designs for new technology
- Discussion around building blocks for design
- Customers want to design around workloads
15:00 Applications dictating system design
- Optimizing infrastructure for customer workloads
- Design goes all the way up the stack
- Software-Defined market
- What do they see in traditional architecture
- Cloud vs. non-cloud, scale-out, architecture
20:00 On-premise cloud architecture
- Everything is being built on cloud architecture
- It’s not just the public cloud
- On-premise cloud
- Budgets are 50-50 public cloud vs. private cloud
- Enterprises what to own their own business
- The public cloud has that convenience factor
25:00 Addressing complexity
- Controlling efficiencies
- Deployments have different benefits for different workloads
- High-value edge deployments
30:00 Serviceability at the edge
- What’s different for edge
- Reliability for edge deployments
- Software is designed to support reliability and uptime
- No silver bullet
- Thermals have an effect
- Physical challenges
- Edge boxes are benefiting from data center designs
35:00 Cooling
- Liquid cooling is more efficient
- There is a greater adoption level for liquid cooling
- Reliability argument
40:00 Latest Vision
- Software v Hardware
- S/W staff has tripled over the past five years
- Software investment is huge
- GPU portfolio is compelling
- Sorting through GPU technology and how it fits into the decision matrix
- What are the options
45:00 Wrap up and closing
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #105: Infrastructure Trends from Edge to Core appeared first on StorageReview.com.
47 min - Podcast #104: The Latest on HPC and Scale-Out Storage
Brian gets Curtis Anderson of Panasas to join him for a deep and detailed view of HPC. Curtis is the Software Architect at Panasas and co-chair of the MLCommons Storage Working Group. What is MLCommons Storage Working Group? MLCommons is an industry consortium working to accelerate machine learning and increase its positive impact on society. It publishes several standards for the performance of AI/ML models, among other things. As one of the co-chairs, Curtis defines standard(s) for evaluating the performance of storage subsystems that feed AI/ML environments.
Brian gets Curtis Anderson of Panasas to join him for a deep and detailed view of HPC. Curtis is the Software Architect at Panasas and co-chair of the MLCommons Storage Working Group. What is MLCommons Storage Working Group? MLCommons is an industry consortium working to accelerate machine learning and increase its positive impact on society. It publishes several standards for the performance of AI/ML models, among other things. As one of the co-chairs, Curtis defines standard(s) for evaluating the performance of storage subsystems that feed AI/ML environments.
As a Software Architect at Panasas, Curtis coordinates all the technology teams in the Sunnyvale office: Platform, OSD (Object Storage Device), CLI/GUI Management, NFS and CIFS gateway, and interfacing to the hardware and QA teams. He also gets to write code again(!).
The market for scale-out storage with a commercial feature set is growing rapidly, and competitors like NetApp have the commercial feature set but haven’t been able to master scale-out. Panasas has 15 years of scale-out performance and the resiliency everyone wants but needs the commercial feature set that some commercial players have.
Brian digs deep into the technology and asks Curtis to explain use cases and the software structure, where Panasas fits in the overall market and their new flash systems.
This is a great podcast; if HPC is of interest, you should give this a listen. It’s only 45 minutes, but if you want to jump around, we have put some timestamps below:
00:00 Intro
- Supercomputing history
- Use cases – HPC
- Hardware
- New features
- Scale-out
- Parallel systems vs. traditional systems
05:00 How Panasas systems work
- Why it was invented
- Best for moderate/large files
- Nodes for workload types
10:00 Software structure
- PanFS
- More on Parallel file systems
- Why didn’t the big providers address the Parallel F/S
15:00 How it handles multiple HPC projects
- Where does Panasas fit in the market
- Media & Entertainment constraints
- Where does PanFS fit
- CGI
20:00 How M&E works
- Why CGI is different
- Difference between M7E and HPC workloads
- What fits in the HPC market
- AI/ML
25:00 Costs
- GPU’s
- Flash expense
- Compute strategies
- Skillset needed
30:00 How to set up a smaller company for HPC
- How to recruit
- Back to the cloud
- Cloud is the honeypot
- Easy to use
- Low cost
- Not set up for HPC
- Can a Panasas system be consumed in the cloud
- How to architect PanFS for Cloud
35:00 Edge Placement
- Hard to justify
40:00 Environmental impact
- Efficiency
- What has Panasas seen as HPC use cases
- Traditional HPC is changing
- Oil and Gas
- Exploration companies are now looking for Salt Fields to store fuel
45:00 How to engage Panasas
Closing and wrap-up
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #104: The Latest on HPC and Scale-Out Storage appeared first on StorageReview.com.
47 min - Podcast #103: A Deep Dive on Immersion Cooling
There has been a lot of buzz around liquid cooling, especially when discussing performance degradation when chips overheat. Liquid cooling technology is a successful solution to keeping components cool and operating at peak performance.
There has been a lot of buzz around liquid cooling, especially when discussing performance degradation when chips overheat. Liquid cooling technology is a successful solution to keeping components cool and operating at peak performance.
StorageReview.com has covered the progress and acceptance of this technology with much focus on GPUs, but it goes deeper. So, while attending Dell Technologies World 2022, Brian was given a chance to look closely at some of the solutions TMGcore had on display. We posted a few short videos from the event, which garnered a lot of conversation. The technology was impressive, so Brian reached out to Taylor Monnig, Chief Operating Officer at TMGcore, to join him for an in-depth discussion on immersion cooling.
TMGcore is focused on developing advanced immersion cooling and has developed a suite of products that includes both hardware and software solutions. Located in Plano, the tech hub of Texas, TMGcore has its fingers in all major technologies like HPC, AI &Machine Learning, Edge, Cloud, Gaming, Government, and more. Taylor takes Brian through liquid cooling options, challenges facing air-cooling, and the paradigm shift to liquid cooling. Taylor also talks about server and card design to support liquid cooling and the differences between the liquids used in the immersion technology.
This is an in-depth discussion with someone who understands and believes in the technology. This emerging technology continues to gain traction, and this podcast provides excellent detail from a company dedicated to immersion cooling.
Social Videos:
- Single-phase demo – TikTok | Instagram
- Two-phase demo – TikTok | Instagram
If you don’t have time to watch the podcast in its entirety, you can jump to the sections that are most interesting to you by checking the timestamps below.
Timestamp
00:00 Introduction
- Introduction to immersion cooling
- Options
- Liquid on-chip
- Rear door heat exchange
- two-phase
- Cooling distribution unit
- Steps to move to immersion
05:00 Challenges
- Air cooling challenges
- HPC
- Paradigm shift to immersion
- Driving migration to liquid cooling
- Design changes to increase performance
10:00 TMGcore Servers
- Server design
- Thermals
- Re-use warm egress water
- What components use immersion
15:00 Impacts
- Impact on power consumption
- Modifications to existing servers
- Eliminating needs for specific components
- Reduce the waste that goes to landfills
- Not for storage
- SSDs
- HDDs
20:00 Servicing
- Device serviceability
- Removing, servicing, re-inserting into liquid
- Hot updates not recommended
- Single-phase vs. two-phase
25:00 Markets
- #1 Crypto
- #2 HPC
- #3 Government
- How two-phase works
- Fluid differences
- Maintenance
30:00 Costs
- Cost of ownership
- Fluid vendors
35:00 Data Center designs
- What are the design considerations for data center layout
- Vision for SMB/Small Enterprise
- What do OEMs need to do to build immersion-ready servers
- Creativity in design
40:00 Edge
- Edge boxes
- Areas of acceptance
- Reasons for acceptance in specific regions
45:00 Deployments
- Rugged design deployments
- Government
- Military
- Foreign deployment
- Logistics
50:00 Wrap up!
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #103: A Deep Dive on Immersion Cooling appeared first on StorageReview.com.
54 min - Podcast #102: Dan Kogan, Product Focus at Pure//Accelerate techfest22
Pure Storage kicked off Pure//Accelerate techfest22 earlier this month in Los Angeles, and that gave Brian a chance to catch up with Dan Kogan, Vice President, Product Management and Product Marketing at Pure Storage. Dan leads product management for new business, cloud, and the as-a-Service group within Pure’s FlashArray BU, along with leading FlashArray’s overall product marketing team.
Pure Storage kicked off Pure//Accelerate techfest22 earlier this month in Los Angeles, and that gave Brian a chance to catch up with Dan Kogan, Vice President, Product Management and Product Marketing at Pure Storage. Dan leads product management for new business, cloud, and the as-a-Service group within Pure’s FlashArray BU, along with leading FlashArray’s overall product marketing team.
Pure unveiled several new and rebranded products during the event. StorageReview.com posted the news as it happened. You can read more about the key announcements for the Expanded Evergreen Portfolio, the FlashBlade//S, and AIRI//S AI-Ready Infrastructure. Pure Storage has been busy, and these announcements are just the beginning.
There is a focus on as-a-Service offerings, strengthening partnerships with NVIDIA, and launching the FlashBlade//S. Dan has his hand in many of the new initiatives within Pure Storage.
Brian has questions about the new announcements, what’s next, cloud deliverables, and more. This very informative podcast offers insights into the direction Pure Storage is taking for the future.
The last few minutes of the video were lost, but the audio was preserved.
Please give it a listen, or skip around using the timestamps below.
00:00 Introduction
- Intro
- What resonates with customers
- Cloud and cloud acceptance
- Datacenters are growing
- It’s complicated
- Hybrid scenarios
- Cloud costs are getting in the way
05:00 Cloud focus
- Applications moving to the cloud: What is the right mix/balance
- Considerations that determine cloud migration
- SMB vs. Enterprise
- Cloud migration is based on company maturity
- Pure Storage customers:
- Large SAS users
- Cloud economics don’t work
10:00 Educating the masses
- Training partners like MSPs, VARS, Integrators
- Pure is not your typical storage vendor
- Remember that Pure is a software company
- The magic sauce is how the software work with the hardware
15:00 Pure//Accelerate techfest22
- What was on display
- Focus on Pure Fusion
- Fusion changes the field for IT’s ability to support and deploy
- Creation of Service Provider modality
20:00 History
- A look back at EMC Viper
- Pure Fusion won’t make the same mistakes
- Pure and cloud – Block Store
- supported on AWS and Azure
25:00 Pricing and Chargeback
- Pricing and chargeback for different tiers can surprise customers
- Pure provides customers with all prices and costs upfront
- Specifics on new and rebranded products
- Emphasis on Evergreen
- New – Evergreen//Flex – subscription service
30:00 Where is everyone going?
- as-a-Service is the choice for vendors
- Pure’s secret sauce – data reduction
- On-premises
- In the cloud
35:00 What are those charges?
- Customers are going all-in on cloud and but the invoice forces a rethink
- Cloud migration and overwhelming options are making cloud harder
- Too many options
- It’s more about as-a-Service
- Portworx discussion
(video lost)
40:00 Portworx
- API level integration
- PDS – Portworx Data Services
- Legacy workloads that can run in a container
- Where will the VMware relationship go with the pending Broadcom purchase
- NVIDIA partnership
- What’s new with FlashBlade//S
- Customers are excited about GPUs
46:00 Wrap-up!
Video pod on YouTube
Subscribe to our podcast:
- iTunes
- Stitcher
- Spotify
- Google Play
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | TikTok | RSS Feed
The post Podcast #102: Dan Kogan, Product Focus at Pure//Accelerate techfest22 appeared first on StorageReview.com.
47 min
About StorageReview.com
From the publisher's feed
StorageReview.com is a leading provider of news and reviews throughout the entire IT stack - from the datacenter to the edge, and all points in between.