Category Archives: Containers

Accelerated Data Paths of High Performance Storage is the Cornerstone of building AI

By | | , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

It has been 2 months into my new role at DDN as a Solutions Architect. With many revolving doors around me, I have been trying to find the essence, the critical cog of the data infrastructure that supports the accelerated computing of the Nvidia GPU clusters. The more I read and engage, a pattern emerged. I found that cog in the supercharged data paths between the storage infrastructure systems and the GPU clusters. I will share more.

To set the context, let me start with a wonderful article I read in CIO.com back in July 2024. It was titled “Storage: The unsung hero of AI deployments“. It was music to my ears because as a long-time practitioner in the storage technology industry, it is time the storage industry gets its credit it deserves.

What is the data path?

To put it simply, a Data Path, from a storage context, is the communication route taken by the data bits between the compute system’s processing and program memory and the storage subsystem. The links and the established sessions can be within the system components such as the PCIe bus or external to the system through the shared networking infrastructure.

High speed accelerated data paths

In the world of accelerated computing such as AI and HPC, there are additional, more advanced technologies to create even faster delivery of the data bits. This is the accelerated data paths between the compute nodes and the storage subsystems. Following on, I share a few of these technologies that are lesser used in the enterprise storage segment.

Continue reading →

Preliminary Data Taxonomy at ingestion. An opportunity for Computational Storage

By | | , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

Data governance has been on my mind a lot lately. With all the incessant talks and hype about Artificial Intelligence, the true value of AI comes from good data. Therefore, it is vital for any organization embarking on their AI journey to have good quality data. And the journey of the lifecycle of data in an organization starts at the point of ingestion, the data source of how data is either created, acquired to be presented up into the processing workflow and data pipelines for AI training and onwards to AI applications.

In biology, taxonomy is the scientific study and practice of naming, defining and classifying biological organisms based on shared characteristics.

And so, begins my argument of meshing these 3 topics together – data ingestion, data taxonomy and with Computational Storage. Here goes my storage punditry.

Data Taxonomy in post-injection 

I see that data, any data, has to arrive at a repository first before they are given meaning, context, specifications. These requirements are different from file permissions, ownerships, ctime and atime timestamps, the content of the ingested data stream are made to fit into the mould of the repository the data is written to. Metadata about the content of the data gives the data meaning, context and most importantly, value as it is used within the data lifecycle. However, the metadata tagging, and preparing the data in the ETL (extract load transform) or the ELT (extract load transform) process are only applied post-ingestion. This data preparation phase, in which data is enriched with content metadata, tagging, taxonomy and classification, is expensive, in term of resources, time and currency.

Elements of a modern event-driven architecture including data ingestion (Credit: Qlik)

Even in the burgeoning times of open table formats (Apache Iceberg, HUDI, Deltalake, et al), open big data file formats (Avro, Parquet) and open data formats (CSV, XML, JSON et.al), the format specifications with added context and meanings are added in and augmented post-injection.

Continue reading →

Deploying a MinIO SNMD Object Storage Server in TrueNAS SCALE

By | | , , , , , , , , , , , , , , , , ,
3 Comments

[ Preamble ] This deployment of MinIO SNMD (single node multi drive) object storage server on TrueNAS® SCALE 24.04 (codename “Dragonfish”) is experimental. I am just deploying this in my home lab for the fun of it. Do not deploy in any production environment.

I have been contemplating this for quite a while. Which MinIO deployment mode on TrueNAS® SCALE should I work on? For one, there are 3 modes – Standalone, SNMD (Single Node Multi Drives) and MNMD (Multi Node Multi Drives). Of course, the ideal lab experiment is MNMD deployment, the MinIO cluster, and I am still experimenting this on my meagre lab resources.

In the end, I decided to implement SNMD since this is, most likely, deployed on top of a TrueNAS® SCALE storage appliance instead an x-86 bare-metal or in a Kubernetes cluster on Linux systems. Incidentally, the concept of MNMD on top of TrueNAS® SCALE is “Kubernetes cluster”-like albeit a different container platform. At the same time, if this is deployed in a TrueNAS® SCALE Enterprise, a dual-controller TrueNAS® storage appliance, it will take care of the “MinIO nodes” availability in its active-passive HA architecture of the appliance. Otherwise, it can be a full MinIO cluster spread and distributed across several TrueNAS storage appliances (minimum 4 nodes in a 2+2 erasure set) in an MNMD deployment scheme.

Ideally, the MNMD deployment should look like this:

MinIO distributed multi-node cluster architecture (credit: MinIO)

Continue reading →

And great AI starts with good Data Management

By | | , , , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

Processing data has become more expensive.

Somewhere, there is a misconception that data processing is cheap. That stems from the well-known pricings of the capacities of public cloud storage that are a fraction of cents per month. But data in storage has to be worked upon, and has to be built up and protected to increase its value. Data has to be processed, moved, shared, and used by applications. Data induce workloads. Nobody keeps data stored forever and never be used again. Nobody buys storage just for capacity alone.

We have a great saying in the industry. No matter, where the data moves, it will land in a storage. So, it is clear that data does not exist in ether. And yet, I often see how little attention and prudence and care, when it comes to data infrastructure and data management technologies, the very components that are foundational to great data.

Great data management for Great AI

AI is driving up costs in data processing

A few recent articles drew my focus into the cost of data processing.

Here is one posted by a friend on Facebook. It is titled “The world is running out of data to feed AI, experts warn.”

My first reaction was, “How can we run out of data“? We have so much data in the world today that the 175 zettabytes predicted by IDC when we reach 2025 might be grossly inaccurate. According to Exploding Topics, it is estimated that we create 328.77TB of data per day, 120 zettabytes per year. While I cannot vouch for the accuracy of the numbers, the numbers are humongous.

Continue reading →

Disaggregation and Composability vital for AI/DL models to scale

By | | , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

New generations of applications and workloads like AI/DL (Artificial Intelligence/Deep Learning), and HPC (High Performance Computing) are breaking the seams of entrenched storage infrastructure models and frameworks. We cannot continue to scale-up or scale-out the storage infrastructure to meet these inundating fluctuating I/O demands. It is time to look at another storage architecture type of infrastructure technology – Composable Infrastructure Architecture.

Infrastructure is changing. The previous staid infrastructure architecture parts of compute, network and storage have long been thrown of the window, precipitated by the rise of x86 server virtualization almost 20 years now. It triggered a tsunami of virtualizing everything, including storage virtualization, which eventually found a more current nomenclature – Software Defined Storage. Both storage virtualization and software defined storage (SDS) are similar and yet different and should be revered through different contexts and similar goals. This Tech Target article laid out both nicely.

As virtualization raged on, converged infrastructure (CI) which evolved into hyperconverged infrastructure (HCI) went fever pitch for a while. Companies like Maxta, Pivot3, Atlantis, are pretty much gone, with HPE® Simplivity and Cisco® Hyperflex occasionally blipped in my radar. In a market that matured very fast, HCI is now dominated by Nutanix™ and VMware®, with smaller Microsoft®, Dell EMC® following them.

From HCI, the attention of virtualization has shifted something more granular, more scalable in containerization. Despite a degree of complexity, containerization is taking agility and scalability to the next level. Kubernetes, Dockers are now mainstay nomenclature of infrastructure engineers and DevOps. So what is driving composable infrastructure? Have we reached the end of virtualization? Not really.

Evolution of infrastructure. Source: IDC

It is just that one part of the infrastructure landscape is changing. This new generation of AI/ML workloads are flipping the coin to the other side of virtualization. As we see the diagram above, IDC brought this mindset change to get us to Think Composability, the next phase of Infrastructure.

Continue reading →

Project COSI

By | | , , , , , , , , , , , , , , , , , ,
Leave a comment

The S3 (Simple Storage Service) has become a de facto standard for accessing object storage. Many vendors claim 100% compatibility to S3, but from what I know, several file storage services integration and validation with the S3 have revealed otherwise. There are certain nuances that have derailed some of the more advanced integrations. I shall not reveal the ones that I know of, but let us use this thought as a basis of our discussion for Project COSI in this blog.

Project COSI high level architecture

What is Project COSI?

COSI stands for Container Object Storage Interface. It is still an alpha stage project in Kubernetes version 1.25 as of September 2022 whilst the latest version of Kubernetes today is version 1.26. To understand the objectives COSI, one must understand the journey and the challenges of persistent storage for containers and Kubernetes.

For me at least, there have been arduous arguments of provisioning a storage repository that keeps the data persistent (and permanent) after containers in a Kubernetes pod have stopped, or replicated to another cluster. And for now, many storage vendors in the industry have settled with the CSI (container storage interface) framework when it comes to data persistence using file-based and block-based storage. You can find a long list of CSI drivers here.

However, you would think that since object storage is the most native storage to containers and Kubernetes pods, there is already a consistent way to accessing object storage services. From the objectives set out by Project COSI, turns out that there isn’t a standard way to provision and accessing object storage as compared to the CSI framework for file-based and block-based storage. So the COSI objectives were set to:

  • Kubernetes Native – Use the Kubernetes API to provision, configure and manage buckets
  • Self Service – A clear delineation between administration and operations (DevOps) to enable self-service capability for DevOps personnel
  • Portability – Vendor neutrality enabled through portability across Kubernetes Clusters and across Object Storage vendors

Further details describing Project COSI can be found here at the Kubernetes site titled “Introducing COSI: Object Storage Management using Kubernetes API“.

Standardization equals technology adoption

Standardization means consistency, control, confidence. The higher the standardization across the storage and containerized apps industry, the higher the adoption of the technology. And given what I have heard from the industry over these few years, Kubernetes, to me, even till this day, is a platform and a framework that are filled and riddled with so many moving parts. Many of the components looks the same, feels the same, and sounds the same, but might not work out the same when deployed.

Therefore, the COSI standardization work is important and critical to grow this burgeoning segment, especially when we are rocketing towards disaggregation of computing service units, resources that be orchestrated to scale up or down at the execution of codes. Infrastructure-as-Code (IAC) is becoming a reality more and more with each passing day, and object storage is at the heart of this transformation for Kubernetes and containers.

Continue reading →

TrueNAS SCALE Clustered SMB to the fore

By | | , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

iXsystems™ released second iteration of TrueNAS® SCALE software just over a week ago. It is known as version 22.02.2 or Anglefish.2, with the most notable upgrades to HA (High Availability) for SCALE and Clustered SMB capabilities. This is the perfect excuse for me to learn about Clustered SMB and share what I have learned.

TrueNAS SCALEFor the

For the uninformed, Clustered SMB brings highly available SMB file sharing services to mission critical environments. More importantly, Clustered SMB is high availability in a scale-out clustered architecture.

My view beyond HA SMB

I am not familiar with Clustered SMB in a NAS (Network Attached Storage). The world I am more familiar with is either having CIFS/SMB file services on a dual controller storage appliance or running Windows File Sharing on an Microsoft® Clustered Service (MSCS). Typically in these 2 types of HA SMB services, the scale up architecture require a shared access to a consolidated storage volume. Behind the scenes, there are many mechanisms at play to ensure that one, and only one, storage controller or HA host can have write access capabilities at one time. The most common mechanism is the SCSI-3 Persistent Reservation or sometimes known as SCSI fencing, using the SPC-3 (SCSI Primary Command) primitives. The whole objective is to prevent 2 nodes or hosts to writing to the shared storage volume at the same time and other issues like split-brain.

Continue reading →

Object Storage becoming storage lingua franca of Edge-Core-Cloud

By | | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

Data Fabric was a big buzzword going back several years. I wrote a piece talking about Data Fabric, mostly NetApp®’s,  almost 7 years ago, which I titled “The Transcendence of Data Fabric“. Regardless of storage brands and technology platforms, and each has its own version and interpretations, one thing holds true. There must be a one layer of Data Singularity. But this is easier said than done.

Fast forward to present. The latest buzzword is Edge-to-Core-Cloud or Cloud-to-Core-Edge. The proliferation of Cloud Computing services, has spawned beyond to multiclouds, superclouds and of course, to Edge Computing. Data is reaching to so many premises everywhere, and like water, data has found its way.

Edge-to-Core-to-Cloud (Gratitude thanks to https://www.techtalkthai.com/dell-technologies-opens-iot-solutions-division-and-introduces-distributed-core-architecture/)

The question on my mind is can we have a single storage platform to serve the Edge-to-Core-to-Cloud paradigm? Is there a storage technology which can be the seamless singularity of data? 7+ years onwards since my Data Fabric blog, The answer is obvious. Object Storage.

The ubiquitous object storage and the S3 access protocol

For a storage technology that was initially labeled “cheap and deep”, object storage has become immensely popular with developers, cloud storage providers and is fast becoming storage repositories for data connectors. I wrote a piece called “All the Sources and Sinks going to Object Storage” over a month back, which aptly articulate how far this technology has come.

But unknown to many (Google NASD and little is found), object storage started its presence in SNIA (it was developed in Carnegie-Mellon University prior to that) in the early 90s, then known as NASD (network attached secure disk). As it is made its way into the ANSI T10 INCITS standards development, it became known as Object-based Storage Device or OSD.

The introduction of object storage services 16+ years ago by Amazon Web Services (AWS) via their Simple Storage Services (S3) further strengthened the march of object storage, solidified its status as a top tier storage platform. It was to AWS’ genius to put the REST API over HTTP/HTTPS with its game changing approach to use CRUD (create, retrieve, update, delete) operations to work with object storage. Hence the S3 protocol, which has become the de facto access protocol to object storage.

Yes, I wrote those 2 blogs 11 and 9 years ago respectively because I saw that object storage technology was a natural fit to the burgeoning new world of storage computing. It has since come true many times over.

Continue reading →

All the Sources and Sinks going to Object Storage

By | | , , , , , , , , , , , , , , , , , , , , , , , , , ,
Leave a comment

The vocabulary of sources and sinks are beginning to appear in the world of data storage as we witness the new addition of data processing frameworks and the applications in this space. I wrote about this in my blog “Rethinking data. processing frameworks systems in real time” a few months ago, introducing my take on this budding new set of I/O characteristics and data ecosystem. I also started learning about the Kappa Architecture (and Lambda as well), a framework designed to craft and develop a set of amalgamated technologies to handle stream processing of a series of data in relation to time.

In Computer Science, sources and sinks are considered external entities that often serve as connectors of input and output of disparate systems. They are often not in the purview of data storage architects. Also often, these sources and sinks are viewed as black boxes, and their inner workings are hidden from the views of the data storage architects.

Diagram from https://developer.here.com/documentation/get-started/dev_guide/shared_content/topics/olp/concepts/pipelines.html

The changing facade of data stream processing presents the constant motion of data, the continuous data being altered as it passes through the many integrated sources and sinks. We are also see much of the data processed in-memory as much as possible. Thus, the data services from a traditional storage model of SAN and NAS may straggle with the requirements demanded by this new generation of data stream processing.

As the world of traditional data storage processing is expanding into data streams processing and vice versa, and the chatter of sources and sinks can no longer be ignored.

Continue reading →

Computational Storage embodies Data Velocity and Locality

By | | , , , , , , , , , , , , , , , , , , ,
1 Comment

I have been earnestly observing the growth of Computational Storage for a number of years now.  It was known by several previous names, with the name “in-situ data processing” stuck with me the most. The Computational Storage nomenclature became more cohesive when SNIA® put together the CMSI (Compute Memory Storage Initiative) some time back. This initiative is where several standards bodies, the major technology players and several SIGs (special interest groups) in SNIA® collaborated to advance Computational Storage segment in the storage technology industry we know of today.

The use cases for Computational Storage are burgeoning, and the functional implementations of Computational Storage are becoming vital to tackle the explosive data tsunami. In 2018 IDC, in its Worldwide Global Datasphere Forecast 2021-2025 report, predicted that the world will have 175 ZB (zettabytes) of data. That number, according to hearsay, has been revised to a heady figure of 250ZB, given the superlative rate data is being originated, spawned and more.

Computational Storage driving factors

If we take the Computer Science definition of in-situ processing, Computational Storage can be distilled as processing data where it resides. In a nutshell, “Bring Compute closer to Storage“. This means that there is a processing unit within the storage subsystem which does not require the host CPU to perform processing. In a very simplistic manner, a RAID card in a storage array can be considered a Computational Storage device because it performs the RAID functions instead of the host CPU. But this new generation of Computational Storage has much more prowess than just the RAID function in a RAID card.

There are many factors in Computational Storage that make a lot sense. Here are a few:

  1. Voluminous data inundate the centralized architecture of the cloud platforms and the enterprise systems today. Much of the data come from end point devices – mobile devices, sensors, IoT, point-of-sales, video cameras, et.al. Pre-processing the data at the origin data points can help filter the data, reduce the size to be processed centrally, and secure the data before they are ingested into the central data processing systems
  2. Real-time processing of the data at the moment the data is received gives the opportunity to create the Velocity of Data Analytics. Much of the data do not need to move to a central data processing system for analysis. Often in use cases like autonomous vehicles, fraud detection, recommendation systems, disaster alerts etc require near instantaneous responses. Performing early data analytics at the data origin point has tremendous advantages.
  3. Moore’s Law is waning. The CPU (central processing unit) is no longer the center of the universe. We are beginning to see CPU offloading technologies to augment the CPU’s duties such as compression, encryption, transcoding and more. SmartNICs, DPUs (data processing units), VPUs (visual processing units), GPUs (graphics processing units), etc have come forth to formulate a new computing paradigm.
  4. Freeing up central resources with Computational Storage also accelerates the overall distributed data processing in the whole data architecture. The CPU and the adjoining memory subsystem are less required to perform context switching caused by I/O interrupts as in most of the compute/storage architecture today. The total effect relieves the CPU and giving back more CPU cycles to perform higher processing tasks, resulting in faster performance overall.
  5. The rise of memory interconnects is enabling a more distributed computing fabric of data processing subsystems. The rising CXL (Compute Express Link™) interconnect protocol, especially after the Gen-Z annex, has emerged a force to be reckoned with. This rise of memory interconnects will likely strengthen the testimony of Computational Storage in the fast approaching future.

Computational Storage Deployment Models

SNIA Computational Storage Universe in 2019

Continue reading →