0101: Greybeards talk with Howard Marks, Technologist Extraordinary & Plenipotentiary at VAST

As most of you know, Howard Marks (@deepstoragenet), Technologist Extraordinary & Plenipotentiary at VAST Data used to be a Greybeards co-host and is still on our roster as a co-host emeritus. When I started to schedule this podcast, it was going to be our 100th podcast and we wanted to invite Howard and the rest of the co-hosts to be on the call to discuss our podcast. But alas, the 100th Greybeards podcast came and went, before we could get it done. So we decided to refocus this podcast back on VAST Data.

We talked with Howard last year about VAST and some of this podcast covers the same ground (see last year’s podcast with Howard on VAST Data) but I highlighted below different aspects of their product that we also discussed.

For starters, VAST just finalized a recent round of funding, which if I recall, valued them at over $1B USD, or yet another data storage unicorn.

VAST is a scale out, disaggregated, unstructured data platform that takes advantage of the economics of QLC SSD (from Intel) combined with the speed of 3D XPoint storage class memory (Optane SSD, also from Intel) to support customer data. Intel is an investor in VAST.

VAST uses mutliple front end (controller) servers, with one or more HA NVMe drive module(s) connected via a dual infiniband or 100Gbps Ethernet RDMA cluster interconnect. The HA NVMe drive module has two (IO modules) adapter cards, one for each connection that takes IO and data requests and transfers them across a PCIe bus which connects to QLC and Optane SSDs. They also have a Mellanox (another investor) switch on their backend with a (round robin) DNS router to connect hosts to their storage (front-end) servers.

Each backend HA NVMe drive module has 12 1.5TB Optane U.2 SSDs and 44 15.4TB QLC SSDs, for a total of 56 drives. Customer data is first written to Optane and then destaged to QLC SSD.

QLC has the advantage of being 4 bits per cell (for a lower $/GB stored) but it’s endurance or drive writes/day (dw/d)) is significantly worse than TLC. So VAST has had to work to increase QLC endurance in their system.

Natively, QLC offers ~0.2 dw/d when doing random 4K writes. However, if your system does 128KB sequential writes, it offers 4.0 dw/d. VAST destages data from Optane SSDs to QLC in 1MB chunks which both optimizes endurance and reduces garbage collection write amplification within the drive.

Howard mentioned their frontend servers are stateless, i.e., maintain no state information about any IO activity going on. Any IO state information is maintained by their system in Optane SSDs. Each server maintains a work log (like) structure on Optane that describes what they are doing in support of host IO and other activities. That way, if one front end server goes down, another one can access its log and take over its activity.

Metadata is also maintained only on Optane SSDs. Howard called their metadata structure a V-tree (B-tree). VAST mirrors all meta-data and customer data to two Optane SSDs. So if one Optane SSD goes down, its pair can be used to continue operations.

In last years podcast we talked at length about VAST data protection and data reduction capabilities so we won’t discuss these any further here.

However, one thing worth noting is that VAST has a very large RAID (erasure code protection) stripe. Data is written to the QLC SSDs in a VAST designed, locally decodable erasure coding format.

One problem with large stripes is rebuild time. VAST’s locally decodable parity codes help with this but the other thing that helps is distributing rebuild IO activity to all front end servers in the system.

The other problem with large stripe sizes is garbage collection. VAST segregates customer data by “temporariness” based on their best guess. In this way all data in one stripe should have similar lifetimes. When it’s time for stripe garbage collection, having all temporary data allows VAST to jettison the whole stripe (or most of it) rather than having to collect and re-write old stripe data to another new stripe.

VAST came out supporting NFSv3 and S3 object storage protocols, Their next release adds support for SMB 2.2, data-at-rest encryption and snapshotting to an external S3 store. As you may recall SMB is a stateful protocol. In VAST’s home grown, SMB implementation, front end servers can take over SMB transactions from other failed servers, without having to fail the whole transaction and start over again.

VAST uses a fail in place, maintenance policy. That is failed SSDs are not normally replaced in customer deployments, rather blocks, pages, or SSDs are marked as failed and the spare capacity available in the drive enclosure is used to provide space for any needed rebuilt data.

VAST offers a 10 year maintenance option where the customer keeps the same storage for 10 full years. That way customers don’t have to migrate data from one system to another until their 10 years are up.

The podcast runs a little under 44 minutes. Howard and I can talk forever. He is always a pleasure to talk with as well as extremely knowledgeable about (VAST) storage and other industry solutions.  The co-hosts and I had a great time talking with him again. Listen to the podcast to learn more.

This image has an empty alt attribute; its file name is Subscribe_on_iTunes_Badge_US-UK_110x40_0824.png
This image has an empty alt attribute; its file name is play_prism_hlock_2x-300x64.png

Howard Marks, Technologist Extraordinary and Plenipotentiary, VAST Data, Inc.

Howard Marks brings over forty years of experience as a technology architect for hire and Industry observer to his role as VAST Data’s Technologist Extraordinary and Plienopotentary. In this role, Howard demystifies VAST’s technologies for customers and customer requirements for VAST’s engineers.

Before joining VAST, Howard ran DeepStorage an industry test lab and analyst firm. An award-winning speaker, he has appeared at events on three continents including Comdex, Interop and VMworld.

Howard is the author of several books (all gratefully out of print) and hundreds of articles since Bill Machrone taught him journalism at PC Magazine in the 1980s.

Listeners may also remember that Howard was a founding co-Host of the Greybeards-on-Storage Podcast.

098: GreyBeards talk data protection & visualization for massive unstructured data repositories with Christian Smith, VP Product at Igneous

Sponsored By:

Even before COVID-19 there was a lot of file data being created and mined, but with the advent of the pandemic, this has accelerated considerably. As such, it seemed an appropriate time to talk with Christian Smith, VP of Product at Igneous, (@IgneousIO) a company that targets the protection and visibility of massive quantities of unstructured data, on premise, in the cloud, or just about anywhere else it may live.

Let me state at the outset, that my belief had always been, that you don’t backup 10PB of data, rather you bite the (big expense) bullet to replicate it and hope for the best. After talking with Christian and Igneous I am going to have to modify that belief by a couple of more orders of magnitude.

All this data is coming from: LIDAR, RADAR, audio, video, pictures, medical film, MRI/CAT Scans, etc., and as noted above, it’s exploding. Christian talked about one customer of theirs that supplies aerial photography/LIDAR/RADAR scans of areas on request. This can used to better understand crop, forest, wildlife, land health and use. One surprise Igneous found with this customer is that the data is typically archived after first use, but within a month or so it’s moved back online for some other purpose.

Igneous heritage

Many of the people who started up and currently work at Igneous have been around file storage for some time having, primarily coming from (Dell EMC) Isilon, NetApp, Qumulo and other industry heavyweights. When they started Igneous, they realized the world didn’t need another NAS box or file system. Rather, with the advent of 10-100PB unstructured data farms, what was needed was an effective way to protect and understand that data.

When they considered how to protect and visualize 100PB of unstructured data, the only they found to do this was to build a scale-out solution that used on premise and cloud infrastructure and was offered as a service.

Igneous DataProtect solution

With 10PB or 100PB of files, located across a gaggle of heterogeneous file servers, with billions of files across ~100s of servers, each of with has ~1K or more file shares, just scanning all the file servers would take weeks, if not longer and then you need to move the data someplace to protect it. Seems like an impossible task.

Igneous immediately figured out the first thing they needed was a radically new, scale out architecture to rapidly scan of the file servers. Thus was born ActiveScan. Christian said it was designed to scan a trillion files and they have customers with a billion files using their service today. ActiveScan doesn’t use NFS/SMB/Object (S3) access protocols to talk with file servers rather it uses internal APIs to access file metadata. DataProtect currently supports APIs for NetApp, Dell EMC Isilon, Pure FlashBlade, Qumulo, Gluster, Lustre, & GPFS (IBM Spectrum Scale) file systems. They use ActiveScan to build a file index database.

Their other major concern was hot to move PBs of data rapidly across to the cloud and other locations. Again they created a scale out, multi-threaded service to do this and also made use of internal APIs rather than standard file or object protocols. This became IntelliMove. That same customer above with billions of files, has 6PB of file data to protect.

Normal data movement is fine for largish, files but bogs down with lots of small files or extremely large files to back up. DataProtect gathers together small files into a large chunks and splits up extremely large files into smaller chunks and moves these chunks to secondary storage.

Data expiration is another problem, especially when you chunk files together. Here they came up with an intelligent garbage collection algorithm which only collects free space when it makes the most sense but deletes data access at the time of expiration.

DataProtect uses a cloud based, SaaS control plane that manages and coordinates its activities across data centers, sites and cloud instances. It also has a client VM (OVA, with 8 core CPU, 32GB DRAM, ~100MB) that runs in the customers infrastructure, on site, in CoLo’s or in the cloud that is used to scan-move-protect customer unstructured data. If more scan and data movement performance is needed, the VM can spawn additional threads automatically and more VMs can be added to provide even more throughput.

DataDiscover solution

The other service that Igneous offers is DataDiscover a data visualization tool. DataDiscover uses ActiveScan and its database to provide customers a way to understand the file data that resides in their massive unstructured data farms across the data center, cloud or wherever else it resides.

We didn’t discuss this solution as much but having a way to better understand the files in a 10-100PB unstructured data farm could be very useful and a great way to keep that 100PB from growing to 1EB faster than it has too.

As part of their outreach to the world, Igneous is giving away free DataProtect services to organizations that are focused on COVID-19 research. Check out their offer here

The podcast ran ~24 minutes. Christian was extremely knowledgeable about the problems that happen with very large unstructured data farms and how Igneous solutions can provide a better way to protect and visualize that data. Matt and I had a fun time discussing Igneous’s approach with Christian. Listen to the podcast to learn more.

This image has an empty alt attribute; its file name is Subscribe_on_iTunes_Badge_US-UK_110x40_0824.png
This image has an empty alt attribute; its file name is play_prism_hlock_2x-300x64.png

Christian Smith, VP Product at Igneous

Christian is VP of Product, responsible for product management, solutions, and customer success. Prior to Igneous, Christian spent 15 years running field engineering organizations at EMC, Isilon Systems, NetApp and Silicon Graphics.

Christian has been working with organizations that work with file data since working at Silicon Graphics. Before that Christian was co-founder of a small management consulting company associated with Y2K and deregulation.

Christian received dual bachelor’s degrees in Chemistry and Computer Science from the University of Missouri-Columbia. Christian is an avid camper, skier and traveler and has long since traveled through all of the continental 48 states.

095: GreyBeards talk file sync&share with S. Azam Ali, VP Customer Success at CentreStack

We haven’t talked with a file synch and share vendor in a while now and Matt was interested in the technology. He had been talking with CentreStack, and found that they had been making some inroads in the enterprise. So we contacted S. Azam Ali, VP of Customer Success at CentreStack and asked if he wanted to talk about their product on our podcast.

File synch and share, is part collaboration tool, part productivity tool. With file synch & share many users share the same files, across many different environments and end point devices. It’s especially popular with road warriors that need access to the same files on the road that reside in corporate data centers. With this technology, files updated anywhere would be available to all.

Most file synch&share systems require you to use their storage. But CentreStack just provides synch and share access to NFS and SMB storage that’s already in the data center.

CentreStack doesn’t use VPNs to access data, many other vendor do. But with CentreStack, one just log’s into a website (with AD credentials) and they have immediate browser access to files.

CentreStack uses a gateway VM, that runs in the corporate data center, configured to share files/file directories/shares. We asked whether they were in the data path and Azam said no. However, the gateway does register for file system notifications (e.g. when files are updated, outside CentreStack, they get notified).

CentreStack does maintain meta-data on the files, directories, shares that are under it’s control. Presumably, once an admin sets it up, it goes out and access the file systems that have shared files and populates their meta-data for those files.

CentreStack works with any NFS and SMB file system as well as NAS servers that support these two. It’s unclear whether customers can have more than one gateway server in their data center supporting synch and share but Azam did say that it wasn’t unusual for customers with multi-data centers to have a gateway in each, to support synch&share requirements for each data center.

They use client software on end point devices, which presents the shared files as an external drive (to Mac), presumably a cloud drive for Windows PCs and similar services (in an App) for other systems (IOS, Android phones, iPad, etc.). We believe Azam said Linux was coming soon.

The client software can be configured in cache mode or offline mode:

  • Cache mode – the admin can configure how much space to use on the endpoint device and the software will cache the most recently used files in that space for faster access
  • Offline mode – the software moves all files that the endpoint login can access, to the device.

In cache mode, when users open a file (not in the most recently used cache), there will be some delay as the system retrieves data from the internet and copies it to the endpoint device. Unclear what the delay might be but it’s probably a function of internet speed and load on the gateway, with possibly some overhead for the NFS/SMB/NAS system to supply the data. If there’s not enough space to hold the file, the oldest non-open file is erased from the cache.

In both modes, Centrestack supports cross domain locking. That is, if one client has a file open (for update), all other systems/endpoints may only access the file in read-only mode. After the file is closed. the file can then be opened for update by other users.

When CentreStack clients are used to update files, the data is stored back in the original file systems with versioning. This way if the data is corrupted, admins can easily return back to a known good copy version.

CentreStack also offers a cloud backup and DR service. Gateway admins can request that synch&share files be backed up to cloud storage (AWS S3, Azure Blob and Wasabi). When CentreStack backups file data to the cloud, it also includes metadata information about the files so they can be re-constituted anywhere.

A CentreStack cloud gateway VM can be activated in the cloud to supply access to backed up files. Unclear whether the CentreStack cloud backup has to be restored to block or file storage first or if it just accesses the data on cloud storage directly. But one customers using CentreStack cloud DR would need to run client software in their applications accessing these files.

Wasabi seemed an odd solution to have on their list of supported cloud storage providers, but Azam said for their market, the economics of Wasabi storage were hard to ignore. See our previous podcast with David Friend, Co-Founder& CEO, Wasabi, to learn more about Wasabi.

CentreStack is licensed on a per user basis, not storage capacity bucking industry trends. But they don’t actually own the storage so it makes sense. For CentreStack cloud backup, customers also have to supply the cloud storage.

They also offer a 30 day free trial on their website with unlimited users. We assume this uses CentreStacks cloud gateway and customers bring their own cloud storage to support it.

The podcast runs about 35 minutes. Azam was a bit more marketing than we are used to, but he warmed up once we started asking questions. Listen to the podcast to learn more.

This image has an empty alt attribute; its file name is Subscribe_on_iTunes_Badge_US-UK_110x40_0824.png
This image has an empty alt attribute; its file name is play_prism_hlock_2x-300x64.png

S. Azam Ali, VP of Customer Success, CentreStack

S. Azam Ali, is VP of Customer Success at CentreStack and is an executive with extensive experience in managing global teams including sales, support and consulting services.

Azam’s channel experience includes on-boarding new partners including creation of marketing and training collateral for the partners. Azam is an executive with a passion for customer success and establishing long term relationships and partnerships.

Azam is also an advisor to startups as well as established technology companies.

85: GreyBeards talk NVMe NAS with Howard Marks, Technologist Extraordinary and Plenipotentiary, VAST Data Inc.

As most of you know, Howard Marks was a founding co-Host of the GreyBeards-On- Storage podcast and has since joined with VAST Data, an NVMe file and object storage vendor headquartered in NY with R&D out of Israel. We first met with VAST at StorageFieldDay18 (SFD18, video presentation). Howard announced his employment at that event. VAST was a bit circumspect at their SFD18 session but Howard seems to be more talkative, so on the podcast we learn a lot more about their solution.

VAST Data is essentially an NFS-S3 object store, scale out solution with both stateless, VAST Data storage servers and JBoF drive enclosures with Optane and NVMe QLC SSDs. Storage servers or JBoFs can be scaled independently. They don’t support tiering or DRAM caching of data but instead seem to use the Optane SSDs as a write buffer for the QLC SSDs.

At the SFD18 event their spokesperson said that they were going to kill off disk storage media. (Ed’s note: Disk shipments fell 18% y/y in 1Q 2019, with enterprise disk shipments at 11.5M units, desktop at 24.5M units and laptops at 37M units).

The hardware

The VAST Data storage servers are in a 2U/4 server configuration, that runs interface protocols (NFS & S3), data reduction (see below), data reformating/buffering etc. They are stateless servers with all the metadata and other control state maintained on JBoF Optane drives.

Each drive enclosure JBoF has 12 Optane SSDs and 44 U.2 QLC (no DRAM/no super cap) SSDs. This means there are no write buffers on the QLC SSDs that can lose data when power failures occur. The interface to the JBoF is NVMeoF, either RDMA-RoCE Ethernet or InfiniBand (customer selected). Their JBoFs have high availability, with dual fabric modules that support 2-100Gbps Ethernet/InfiniBand ports per module, 4 per JBoF.

Minimum starting capacity is 500TB and they claim support up to Exabytes. Although how much has actually been tested is an open question. They also support billions of objects/files.

Guaranteed better data reduction

They have a rather unique, multi-level, data reduction scheme. At the start, data is chunked in variable length chunks. They use heuristics to determine the chunk size that fits best. (Ed note, unclear which is first in this sequence below so presented in (our view of) logical order)

  • 1st level computes a similarity hash (56 bit not SHA1), which is used to determine a similarity level with any other currently stored data chunk in the system.
  • 2nd level uses a ZSTD compression algorithm. If a similarity is found, the new data chunk is compressed with the ZSTD compression algorithm and a reference dictionary used by the earlier, similar data chunk. If no existing chunk is similar to this one, the algorithm identifies a semi-unique reference dictionary that optimizes the compression of this data chunk. This semi-unique dictionary is stored as metadata.
  • 3rd level, If it turns out to be a complete duplicate data chunk, then the dedupe count for the original data chunk is incremented, a pointer is saved to the original unique data and the data discarded. If not a complete duplicate of other data, the system computes a delta from the closest “similar’ block and stores just the delta bytes, includes a pointer to the original similar block and increments a delta block counter.

So data is chunked, compressed with a optimized dictionary, be delta-diffed or deduped. All data reduction is done post data write (after the client is ACKed), and presumably, re-hydrated after being read from SSD media. VAST Data guarantees better data reduction for your stored data than any other storage solution.

New data protection

They also supply a unique Locally Decodable Erasure Coding with 4 parity (-like) blocks and anywhere from 36 (single enclosure leaving 4 spare u.2 SSDs) to 150 data blocks per stripe all of which support up to 4 device failures per stripe. 

The locally decodable erasure coding scheme allows for rebuilds without having to read all remaining data blocks in a stripe. In this scheme, once you read the 4 parity (-like) blocks, one has all the information calculated from up to ¾ of the remaining drives in the stripe, so the system only has to read the remaining ¼ drives in the stripe to reconstruct one, two, three, or four failing drives.  Given their data stripe width, this cuts down on the amount of data needing to be read considerably. Still with 150 data drives in a stripe, the system still has to read 38 drives worth of QLC SSD data to rebuild a data drive.

In addition to all the above, VAST Data also reblocks the data into much larger segments, (it writes 1MB segments to the QLC drives) and uses a heat map along with other heuristics to separate actively written data from less actively written data, thus reducing garbage collection, write amplification.

The podcast is a long and runs over ~43 minutes. Howard has always been great to talk with and if anything, now being a vendor, has intensified this tendency. Listen to the podcast to learn more.

Howard Marks, Technologist Extraordinary and Plenipotentiary, VAST Data, Inc.

Howard Marks brings over forty years of experience as a technology architect for hire and Industry observer to his role as VAST Data’s Technologist Extraordinary and Plienopotentary. In this role, Howard demystifies VAST’s technologies for customers and customer requirements for VAST’s engineers.

Before joining VAST, Howard ran DeepStorage an industry test lab and analyst firm. An award-winning speaker, he has appeared at events on three continents including Comdex, Interop and VMworld.

Howard is the author of several books (all gratefully out of print) and hundreds of articles since Bill Machrone taught him journalism at PC Magazine in the 1980s.

Listeners may also remember that Howard was a founding co-Host of the Greybeards-on-Storage Podcast.


60: GreyBeards talk cloud data services with Eiki Hrafnsson, Technical Director, NetApp

Sponsored by:In this episode, we talk with Eiki Hraffnsson (@Eirikurh), Technical Director, NetApp Cloud Data Services.  Eiki gave a great talk at Cloud Field Day 3 (CFD3), although neither Howard nor I were in attendance. I just met Eiki at a NetApp Spring Analyst event earlier this month and after that Howard and I had a chance to talk with him about what’s new in NetApp Cloud Data Services

This is the fourth time NetApp has been on our show (see our podcast with Lee Caswell and Dave Wright,  podcast with Andy Banta, & last month’s sponsored podcast with Adam Carter) and this is their second sponsored podcast.

Eiki came from a company NetApp acquired last year called GreenQloud whose product was QStack. Since then, QStack has become an integral part of their Cloud Data Services.

NetApp has a number of solutions under their Cloud Data Services umbrella and his area of specialty is NetApp Cloud Data Volumes, soon to be available in the MarketPlace on AWS, already in public preview an Microsoft Azure Enterprise NFS and as of 7 May 2018, in private preview as NetApp Cloud Volumes for Google Cloud Platform.

NetApp Cloud Data Volumes

NetApp’s Cloud Data Volume is a public cloud based, storage-as-a-service that supplies enterprise class NFS and SMB (CIFS) storage on a pay as you go model for major public cloud providers. That way your compute instances can have access to predictable performance, highly available file storage in the  cloud.

One advantage that Cloud Data Volumes adds to the public cloud is performance SLAs. That is customers can purchase Low, Medium and High performance file storage. Eiki said they measured Cloud Data Volume IO performance and it achieved almost 10X the public cloud normal (file) storage performance. I assume this was HIGH performing Cloud Data Volume storage, and no information on which storage type was used as the cloud alternative.

Cloud Data Volume customers also get access to NetApp Snapshot services which can create, space efficient, quick read-only copies of their cloud file storage. Cloud Data Volume storage can be purchased on a $/GB/month basis. Other  purchase options are also available for customers who prefer a pre billed amount rather than a consumptive model.

Behind the scenes, Cloud Data Volumes is actually NetApp ONTAP storage. They won’t say what kind or how much, but they do say that NetApp storage is located in public cloud data centers and is fully managed by NetApp.

Customers can use the public cloud native services portal to purchase Cloud Data Volume storage (for Microsoft Azure and GCP) or the NetApp Cloud web portal (for AWS). Once purchased, customers can use an extensive set of native cloud APIs to provision, access and tear-down Cloud Volume storage.

Other NetApp Cloud Data Services

Eiki mentioned that Cloud Data Volumes is just one of many offerings from NetApp’s Cloud Data Services business unit, including:

  • NetApp Private Storage– colocated NetApp storage owned by customers that is adjacent to public clouds.
  • ONTAP Cloud – software defined ONTAP storage system that run in the cloud on compute services using cloud storage to provide block storage.
  • Cloud Sync – data synchronization as a service offering used to replicate data from onprem NAS and object storage to the public cloud.

Probably a few others I am missing here and my bet is more offerings are on the way.

Another item Eiki mentioned with the open source,  NetApp Trident Plugin (GitHub repo). Containers are starting to need persistent state information and this means they need access to storage.

Trident provides dynamic, API driven provisioning of storage volumes for containers under Kubernetes.  Container developers define environmental characteristics which dictate operational environment and now with Trident, can also specify needed storage volumes. That way, when Kubernetes fires up a container for execution, NetApp storage is provisioned just-in-time to support container stateful execution.

The podcast runs ~25 minutes. Eiki was very knowledgeable and was easy to talk with especially on cloud technologies and how NetApp fits in.  Listen to the podcast to learn more.

Erikur (Eiki) Hrafnsson, Technical Director, NetApp Cloud Data Services

Erikur (Eiki) Hrafnsson is an entrepreneur, dad, singer. founder of GreenQloud and maker of QStack, the hybrid cloud platform, now part of NetApp Cloud Data Services. Eiki brings deep public cloud integration knowledge and broad experience in cloud automation and APIs.